SELF-DRIVING IN AN THE HOME MICRO- THE DRONE AND HOW TO HALT

AUGMENTED REALITY GRID IN A BOX THE WHALE ROBOCALLS

How AR will make Single-home grids Flying SnotBots keep Using data to stop
driverless cars safer could unleash solar tabs on ocean health the scammers
P. 22 P. 28 P. 40 P.48

FOR THE TECHNOLOGY INSIDER | 12.19

ATLAS EXCLUSIVE
ACTION
SHRUGS…
AND RUNS,
PHOTOS
OF BOSTON
DYNAMICS’
AND JUMPS, DAZZLING
AND LEAPS, HUMANOID
AND FLIPS… P. 34
Challenge us.
What are your requirements for arbitrary waveform generation?
Our mission: providing instrumentation, software, and advice
to get you the best results with minimal time and effort.

HDAWG
Arbitrary Waveform Generator
starting at
∏Up to 144 synchronized channels $29,325
∏750 MHz, 16 bit, 2.4 GSa/s,
5 Vpp maximum amplitude
∏LabOne® AWG Sequencer and Compiler
∏Digital modulation at multiple frequencies
∏ NEW Real-time precompensation

Zurich Find out more today

Instruments www.zhinst.com
 CONTENTS_12.19

HOW TO SHUT DOWN ROBOCALLERS

STIR/SHAKEN is here to help anyone tired of robocalls. Page 46
by Jim M c Eachern & Eric Burger

22 AUGMENTED REALITY
FOR ROBOCARS
A thorough test must mix real cars The Institute
with simulated ones. TI-6 SHOWCASING
By Yiheng Feng &
Henry X. Liu STARTUPS
IEEE members
share their path to
28 MICROGRID IN A BOX entrepreneurship.
Want rooftop solar to be its best self
for homes and utilities? Storage
and a microgrid will do the job.
By Scott Hinson
IEEE SPECTRUM
(ISSN 0018-9235) is published monthly by The
Institute of Electrical and Electronics Engineers,
Inc. All rights reserved. © 2019 by The Institute
34 BY LEAPS AND BOUNDS of Electrical and Electronics Engineers, Inc.,
3 Park Avenue, New York, NY 10016-5997,
How Boston Dynamics is taking U.S.A. Volume No. 56, Issue No. 12. The editorial
robot agility to new heights. content of IEEE Spectrum magazine does not
represent official positions of the IEEE or its
By Erico Guizzo organizational units. Canadian Post International
Photography by Bob O’Connor Publications Mail (Canadian Distribution) Sales
Agreement No. 40013087. Return undeliverable
Canadian addresses to: Circulation Department,
IEEE Spectrum, Box 1051, Fort Erie, ON L2A
6C7. Cable address: ITRIPLEE. Fax: +1 212 419
7570. INTERNET: spectrum@ieee.org. ANNUAL
40 SNOTBOT: A WHALE OF A SUBSCRIPTIONS: IEEE Members: $21.40
DEEP-LEARNING PROJECT included in dues. Libraries/institutions: $399.
POSTMASTER: Please send address changes
Drones plus machine learning to IEEE Spectrum, c/o Coding Department,
help researchers track the health IEEE Service Center, 445 Hoes Lane, Box 1331,
Piscataway, NJ 08855. Periodicals postage
of whales. paid at New York, NY, and additional mailing
By Bryn Keller & Ted Willke offices. Canadian GST #125634188. Printed at
120 Donnelley Dr., Glasgow, KY 42141-1060,
U.S.A. IEEE Spectrum circulation is audited by
BPA Worldwide. IEEE Spectrum is a member
of the Association of Business Information &
Media Companies, the Association of Magazine
07 NEWS Media, and Association Media & Publishing.
14 RESOURCES IEEE prohibits discrimination, harassment, and
bullying. For more information, visit https://
20 OPINION www.ieee.org/web/aboutus/whatis/policies/
54 PAST FORWARD p9-26.html.

On the cover
Photograph for IEEE Spectrum
by Bob O’Connor

PHOTOGRAPH BY Dan Saelinger SPECTRUM.IEEE.ORG | DEC 2019 |  01

 BACK STORY_

LIGHTS, CAMERA, ROBOTS!

O
ne day this past August, famed robot maker Boston
Dynamics granted IEEE Spectrum access to its highly
restricted headquarters in Waltham, Mass., for an exclusive
photo shoot. Things started off smoothly that morning, but
then one of the robots seemed determined to rush out of the
lab. Atlas, a 1.5-meter-tall humanoid robot, ran a bit farther
than it was supposed to, slammed into a plastic safety barrier,
and fell on its back.
When SWaP Yikes! Did we just participate in the destruction of an unbelievably
costly android?
matters most - Atlas, er, shrugged it off. An instant later, it was back on its feet,
jogging and jumping in front of our cameras, a demonstration of how
Connect with far Boston Dynamics has come in developing some of the world’s
Confidence most agile and robust robots.
In this issue, we bring you unique images of Atlas [p. 34], whose
athletic stunts draw millions of views on YouTube, and Spot, a
Mixed Layout
versatile quadruped that Boston Dynamics is now commercializing.
Miniature Connectors You can also see the photos in an interactive format on our Robots
with 10A per Contact Guide website: https://robots.ieee.org.
For the shoot, we partnered with two Boston-area photographers
who have plenty of expertise with robots. Carlton SooHoo [above,
For data and right] was tasked with capturing 360-degree views of Atlas and Spot
samples go to: using a spinning platform. The trickiest part was keeping Spot still.
www.harwin.com/ The robot reacted to even the slightest vibrations of the platform.
gecko-mt “It was like dealing with a living creature in some ways,” SooHoo says.
Our second photographer, Bob O’Connor, had an even more
challenging job: Capturing freeze-frames of Atlas and Spot in action.
To do that, he set his camera to shoot 14 frames per second, syncing
it to a Profoto Pro-10, a specialized power pack for strobe lights. The
setup guaranteed that the photos looked sharp, with no motion blur.
“To see Atlas run around in the large space we had, fall down, get
itself back up, and walk back to the starting point on its own was
pretty amazing,” O’Connor says.
12.19

By late afternoon, the photographers had amassed a pile of stunning

photos. It was time to call it a wrap. And maybe also breathe a sigh of
relief: No robots were harmed in the making of this shoot. ■

PHOTOGRAPH BY Simon Simard

 L O O K
D E E P E R

Unbreakable materials
lead to unbreakable bonds.
At UC San Diego’s CaliBaja Center,
engineering professor and director
Olivia Graeve has brought together
some of the biggest brains in the U.S.
and Mexico to tackle the smallest
of engineering challenges: developing
hyper-durable nanoparticles. But she doesn’t
stop there. Here, at the No. 1 university
in the nation for women in STEM, she’s
uniting students from both countries in
a summer research program. Proof that
artificial borders can’t hold back
the power of a common goal.

lookdeeper.ucsd.edu

BestColleges.com, 2016
 CONTRIBUTORS_
Scott Hinson
Hinson is the chief technology officer at Pecan
Street, a nonprofit research group in Austin,
Texas, that does data collection and field trials
to accelerate renewables deployment and
conservation. In this issue he writes about a
residential microgrid that Pecan Street developed
[p. 28]. With such a system, Hinson says, utilities
could add more residential solar without worrying
about declines in the quality of power. “They
wouldn’t have any reason to limit solar,” he says.
Bryn Keller
Keller is a deep-learning research scientist in
Intel’s Brain-Inspired Computing Lab. The group,
under the direction of senior principal engineer
Ted Willke, uses ideas from neuroscience to
consider the future of computing. Keller once
dreamed of being a fighter pilot. While that
dream never came to pass, he is now a certified
drone pilot, thanks to the research he and Willke
write about in “SnotBot: A Whale of a Deep-
Learning Project” [p. 40].
Henry X. Liu
Liu is a civil engineering professor at the
University of Michigan. Validating the safety
of autonomous vehicles would take millions
of kilometers of road driving, Liu says. And it’s
hard to simulate the hardest problems, or “edge
cases,” on a test track. He leads a project that uses
augmented reality to safely model even the most
dangerous edge cases. He and coauthor Yiheng
Feng, also of the University of Michigan, tell the
story in this issue [p. 22].
Jim McEachern
Why do we have filters for email spam but no
similar bulwark against unwanted robocalls?
McEachern, the principal technologist at the
telecom industry association ATIS, says the
main problem has been developing a way for
phones—of all things—to communicate properly.
In “How to Shut Down Robocallers” [p. 46], he
and Eric Burger, a computer science researcher at
Georgetown University, describe new protocols
that should stop the flood of unwanted calls.
Sandy Ong
Ong is a science journalist in Singapore. When an
air taxi called Volocopter (made by a company
of the same name) took its maiden flight there
on a rainy day this past October, Ong was on
hand to watch [p. 7]. Volocopter CEO Florian
Reuter said ahead of the event that he looked
forward to “just seeing people’s reactions.”
Meanwhile, Ong’s challenge was to keep her
reporting grounded in the very real obstacles
that air taxis must yet overcome.
04  | DEC 2019 | SPECTRUM.IEEE.ORG
EDITOR IN CHIEF Susan Hassler, s.hassler@ieee.org
EXECUTIVE EDITOR Glenn Zorpette, g.zorpette@ieee.org
EDITORIAL DIRECTOR, DIGITAL
Harry Goldstein, h.goldstein@ieee.org
MANAGING EDITOR Elizabeth A. Bretz, e.bretz@ieee.org
SENIOR ART DIRECTOR
Mark Montgomery, m.montgomery@ieee.org
SENIOR EDITORS
Stephen Cass (Resources), cass.s@ieee.org
Erico Guizzo (Digital), e.guizzo@ieee.org
Jean Kumagai, j.kumagai@ieee.org
Samuel K. Moore, s.k.moore@ieee.org
Tekla S. Perry, t.perry@ieee.org
Philip E. Ross, p.ross@ieee.org
David Schneider, d.a.schneider@ieee.org
Eliza Strickland, e.strickland@ieee.org
DEPUTY ART DIRECTOR Brandon Palacio, b.palacio@ieee.org
PHOTOGRAPHY DIRECTOR Randi Klett, randi.klett@ieee.org
ONLINE ART DIRECTOR Erik Vrielink, e.vrielink@ieee.org
NEWS MANAGER Amy Nordrum, a.nordrum@ieee.org
ASSOCIATE EDITORS
Willie D. Jones (Digital), w.jones@ieee.org
Michael Koziol, m.koziol@ieee.org
SENIOR COPY EDITOR Joseph N. Levine, j.levine@ieee.org
COPY EDITOR Michele Kogon, m.kogon@ieee.org
EDITORIAL RESEARCHER Alan Gardner, a.gardner@ieee.org
ADMINISTRATIVE ASSISTANT
Ramona L. Foster, r.foster@ieee.org
PHOTOGRAPHY INTERN Erica Snyder
CONTRIBUTING EDITORS Evan Ackerman, Mark Anderson,
Robert N. Charette, Peter Fairley, W. Wayt Gibbs, Tam Harbert,
Mark Harris, David Kushner, Robert W. Lucky, Prachi Patel,
Morgen E. Peck, Richard Stevenson, Lawrence Ulrich
EDITOR IN CHIEF, THE INSTITUTE
Kathy Pretz, k.pretz@ieee.org
ASSISTANT EDITOR, THE INSTITUTE
Joanna Goodrich, j.goodrich@ieee.org
DIRECTOR, PERIODICALS PRODUCTION SERVICES Peter Tuohy
EDITORIAL & WEB PRODUCTION MANAGER Roy Carubia
SENIOR ELECTRONIC LAYOUT SPECIALIST Bonnie Nani
PRODUCT MANAGER, DIGITAL Shannan Dunlap
WEB PRODUCTION COORDINATOR Jacqueline L. Parker
MULTIMEDIA PRODUCTION SPECIALIST Michael Spector
IEEE  BOARD OF DIRECTORS
PRESIDENT & CEO José M.F. Moura, president@ieee.org
+1 732 562 3928 FAX: +1 732 465 6444
PRESIDENT-ELECT Toshio Fukuda
TREASURER Joseph V. Lillie SECRETARY Kathleen A. Kramer
PAST PRESIDENT James A. Jefferies
VICE PRESIDENTS
Witold M. Kinsner, Educational Activities; Hulya Kirkici,
Publication Services & Products; Francis B. Grosz Jr., Member
& Geographic Activities; K.J. “Ray” Liu, Technical Activities;
Robert S. Fish, President, Standards Association; Thomas M.
Coughlin, President, IEEE-USA
DIVISION DIRECTORS
Renuka P. Jindal (I); David B. Durocher (II); Sergio Benedetto
(III); John P. Verboncoeur (IV); John W. Walz (V); Manuel Castro
(VI); Bruno Meyer (VII); Elizabeth L. “Liz” Burd (VIII); Alejandro
“Alex” Acero (IX); Ljiljana Trajkovic (X)
REGION DIRECTORS
Babak Dastgheib-Beheshti (1); Wolfram Bettermann (2);
Gregg L. Vaughn (3); David Alan Koehler (4); Robert C.
Shapiro (5);  Keith A. Moore (6); Maike Luiken (7);
Magdalena Salazar-Palma (8); Teófilo J. Ramos (9);
Akinori Nishihara (10)
DIRECTOR EMERITUS Theodore W. Hissey
IEEE STAFF
EXECUTIVE DIRECTOR & COO Stephen Welby
+1 732 562 5400, s.p.welby@ieee.org
CHIEF INFORMATION OFFICER Cherif Amirat
+1 732 562 6017, c.amirat@ieee.org
PUBLICATIONS Michael B. Forster
+1 732 562 3998, m.b.forster@ieee.org
CHIEF MARKETING OFFICER Karen L. Hawkins
+1 732 562 3964, k.hawkins@ieee.org
ADVERTISING PRODUCTION +1 732 562 6334
ADVERTISING PRODUCTION MANAGER
Felicia Spagnoli, f.spagnoli@ieee.org
SENIOR ADVERTISING PRODUCTION COORDINATOR
Nicole Evans Gyimah, n.gyimah@ieee.org
EDITORIAL ADVISORY BOARD, IEEE SPECTRUM
Susan Hassler, Chair; Steve Blank, David C. Brock, Sudhir Dixit,
Shahin Farshchi, Limor Fried, Robert Hebner, Jason K. Hui,
Grant Jacoby, Leah Jamieson, Mary Lou Jepsen, Deepa Kundur,
Norberto Lerendegui, Steve Mann, Allison Marsh, Sofia Olhede,
Jacob Østergaard, Umit Ozguner, John Rogers,
Jonathan Rothberg, Umar Saif, Takao Someya,
Maurizio Vecchione, Yu Zheng, Kun Zhou, Edward Zyszkowski
EDITORIAL ADVISORY BOARD, THE INSTITUTE
Kathy Pretz, Chair; Qusi Alqarqaz, John Baillieul, Philip Chen,
Shashank Gaur, Susan Hassler, Hulya Kirkici, Cecilia Metra,
San Murugesan, Mirela Sechi Annoni Notare, Joel Trussell,
Hon K. Tsang, Chonggang Wang
MANAGING DIRECTOR, PUBLICATIONS
Michael B. Forster
EDITORIAL CORRESPONDENCE
IEEE Spectrum, 3 Park Ave., 17th Floor,
New York, NY 10016-5997
TEL: +1 212 419 7555 FAX: +1 212 419 7570
BUREAU Palo Alto, Calif.; Tekla S. Perry +1 650 752 6661
DIRECTOR, BUSINESS DEVELOPMENT,
MEDIA & ADVERTISING Mark David, m.david@ieee.org
ADVERTISING INQUIRIESNaylor Association Solutions,
Erik Henson +1 352 333 3443, ehenson@naylor.com
REPRINT SALES +1 212 221 9595, ext. 319
REPRINT PERMISSION / LIBRARIES Articles may be
photocopied for private use of patrons. A per-copy fee must
be paid to the Copyright Clearance Center, 29 Congress
St., Salem, MA 01970. For other copying or republication,
contact Managing Editor, IEEE Spectrum.
COPYRIGHTS AND TRADEMARKS IEEE Spectrum is a
registered trademark owned by The Institute of Electrical and
Electronics Engineers Inc. Responsibility for the substance
of articles rests upon the authors, not IEEE, its organizational
units, or its members. Articles do not represent official
positions of IEEE. Readers may post comments online;
comments may be excerpted for publication. IEEE reserves
the right to reject any advertising.
CORPORATE ACTIVITIES Donna Hourican
+1 732 562 6330, d.hourican@ieee.org
MEMBER & GEOGRAPHIC ACTIVITIES Cecelia Jankowski
+1 732 562 5504, c.jankowski@ieee.org
STANDARDS ACTIVITIES Konstantinos Karachalios
+1 732 562 3820, constantin@ieee.org
EDUCATIONAL ACTIVITIES Jamie Moesch
+1 732 562 5514, j.moesch@ieee.org
GENERAL COUNSEL & CHIEF COMPLIANCE OFFICER
Sophia A. Muirhead +1 212 705 8950, s.muirhead@ieee.org
CHIEF FINANCIAL OFFICER &
ACTING CHIEF HUMAN RESOURCES OFFICER
Thomas R. Siegert +1 732 562 6843, t.siegert@ieee.org
TECHNICAL ACTIVITIES Mary Ward-Callan
+1 732 562 3850, m.ward-callan@ieee.org
MANAGING DIRECTOR, IEEE-USA Chris Brantley
+1 202 530 8349, c.brantley@ieee.org
IEEE  PUBLICATION SERVICES & PRODUCTS BOARD
Hulya Kirkici, Chair; Derek Abbott, Petru Andrei,
John Baillieul, Sergio Benedetto, Ian V. “Vaughan” Clarkson,
Eddie Custovic, Samir M. El-Ghazaly, Ron B. Goldfarb,
Larry Hall, Ekram Hossain, W. Clem Karl, Ahmed Kishk,
Aleksandar Mastilovic, Carmen S. Menoni, Paolo Montuschi,
Lloyd A. “Pete” Morley, George Ponchak, Annette Reilly,
Sorel Reisman, Gianluca Setti, Gaurav Sharma, Maria Elena
Valcher, John Vig, Steve Yurkovich, Bin Zhao, Reza Zoughi
IEEE OPERATIONS CENTER
445 Hoes Lane, Box 1331
Piscataway, NJ 08854-1331 U.S.A.
Tel: +1 732 981 0060 Fax: +1 732 981 1721
Smartphones, smart homes, smart...healthcare?

Visualization of the electric field norm and

far-field radiation pattern of a UHF RFID tag.

RFID tags are used across many industries, but when it comes
to healthcare, there is a major design challenge: size. If wearable
RFID tags are too big and bulky, they could cause patient
discomfort. Or, if the tag is for a biomedical implant, it has to be
smaller than a grain of rice! Design engineers can optimize the
size of an RFID tag for its intended purpose using RF simulation.
The COMSOL Multiphysics® software is used for simulating
designs, devices, and processes in all fields of engineering,
manufacturing, and scientific research. See how you can apply
it to designing RFID tags.
comsol.blog/biomed-RFID-tags
 SPECTRAL LINES_
AI-Enabled Spectrum
Technology: What’s Next?
DARPA challenge reveals that AI-managed
spectrum sharing still has obstacles to overcome
Y
ou’ve graduated from the school of spectral hard knocks,” Paul
Tilghman, a U.S. Defense Advanced Research Projects Agency
program manager, told the teams competing in DARPA’s Spec-
trum Collaboration Challenge (SC2) finale on 23 October. The
three-year competition had just concluded. ¶ “Hard knocks”
wasn’t an exaggeration—the 10 teams that made it to the finale,
as well as others who were eliminated in earlier rounds of the
competition, had been tasked with doing something that had
never been done before. Their challenge was to see if AI-managed radio
systems could work together to share wireless spectrum more effectively
than static, preallocated bands. They had spent three years battling it out
in matchups in Colosseum, an RF emulator that DARPA built specially
for the competition. ¶ By the end, the top teams had demonstrated that
their systems could transmit more data over less spectrum than is pos-
sible using existing standards like LTE; they also showed an impressive
ability to reuse spectrum over multiple radios. In some matchups, the
radio systems of five teams were transmitting 200 or 300 percent more
data than is possible with today’s rigid spectrum band allocations. And
that’s important, given that we’re facing a looming wireless-spectrum
crunch. ¶ But, as T ­ ilghman also stressed during SC2, when a DARPA
Grand Challenge ends, it doesn’t mean the technology is ready to go to
market. These challenges are more about proving that a new technologi-
cal idea is possible. So the contest showed that AI-controlled radios can
work together to share spectrum among themselves, and pack more data
into a given amount of spectrum. But where does the technology go from
here, now that the DARPA challenge is finished?
06  | DEC 2019 | SPECTRUM.IEEE.ORG
12.19
At MWC Los Angeles, the mobile-­
industry convention at which the final
competition took place, a group made
up of industry experts, employees of the
U.S. Department of Defense and the U.S.
Federal Communications Commission,
and members of the top three teams all
weighed in. They agreed that there’s still
room for more collaboration.
“Now that the competition is over,”
­T ilghman said, “I know many of the
teams are looking forward to working
together with their competitors.”
The competitive nature of the chal-
lenge meant that teams didn’t necessarily
build their systems to be as truly collab-
orative and altruistic as they may need
to be in the real world. GatorWings’ and
third-place team Zylinium’s systems both
had modes that selfishly hoarded spec-
trum from other well-performing teams.
There are also larger technical ques-
tions that must be resolved. Members of
the GatorWings team talked about how
they spent two years building the cus-
tom software-defined radios for their sys-
tem—and these team members each have 20 years
of experience in RF engineering. The winning teams
agreed that there’s still plenty of work to be done in
refining the radios, including improving the systems’
“brawn”—a measure of how well their signals can
resist interference and still get their own data through.
Julius Knapp, chief of the FCC’s Office of Engineer-
ing and Technology, said it could be tricky to get
competing service providers to share spectrum in
the real world, as collaborative radio efforts require
sharing information about customers that compa-
nies might not be keen to disclose.
For at least one opportunity, though, the final
competition of SC2 seems well timed: the deploy-
ment of 5G networks. Several bands of spectrum
are being developed for this next generation of wire-
less—including, for the first time, millimeter-wave
spectrum. But 5G also relies on immensely valuable
midband spectrum, which remains coveted and in
short supply. Developing collaborative spectrum
technologies alongside 5G could make the wireless
generation more successful. —Michael Koziol
A version of this article appears in our Tech Talk blog.
DARPA
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/sharespectrum1219
 ELECTRIC AIR
TAXI FLIES OVER
SINGAPORE 
Volocopter conducts first test flight
in an urban environment

Beneath a rainy October sky, Sin-

gapore’s business district looked
listless. The glass skyscrapers didn’t glit-
ter and no sunlight dappled in the bay.
But that didn’t matter much because
the crowd had come to gawk at some-
thing else.
At noon, from a promontory across
the bay, a speck of white rose into the air.
With a lawn-mower-like hum, an elec-
tric taxi approached, drawing a swell
of cheers.
Volocopter’s 3-minute test flight was
not the first time the German aircraft
manufacturer has flown its full-scale
prototype publicly. But the demonstra-
tion marks its first test flight in Asia and
the first time the aircraft operated in an
urban environment.
“In the next 10 years, we hope to see GOOD NEIGHBORS? Each Volocopter has raised close to US $95 million. That
Volocopter integrated as an addition to 18 rotors, which the company claims give the air cash has allowed Volocopter to present
taxis a “pleasant sound signature.”
existing mobility methods in megacities,” its third generation of lithium-battery-­
says Christian Bauer, who is in charge powered, two-seater air taxis. Its next
of the firm’s business development. oped worldwide. But only a handful of prototype, VoloCity, will launch by
NIKOLAY KAZAKOV/VOLOCOPTER

­Volocopter aims to be the first company Volocopter’s competitors have actually 2022 and have an estimated range of
in the world to offer commercial electric- flown prototypes. 35 kilometers and a top speed of close
air-taxi services to the masses. Volocopter, which was founded in to 110 kilometers per hour.
More than 215 electric air taxis, also 2011 and counts Daimler, Intel, and the The Volocopter in Singapore was flown
called electric vertical takeoff and land- ­Z hejiang Geely Holding Group (which by a pilot, and Volocopter’s first stage
ing (eVTOL) aircraft, are being devel- owns Volvo) among its investors, has of commercial operations, scheduled

SPECTRUM.IEEE.ORG | DEC 2019 |  07

to happen within five years, will
likely involve piloted flights. But
the company’s ultimate vision is
for its electric air taxis to operate
autonomously. In 2017, the com-
pany completed an uncrewed test
flight in Dubai.
“Volocopter is focused on serving
the inner-city mission,” says CEO
Florian Reuter. With one-way fares
in the “hundreds rather than thou-
sands of dollars,” Reuter says the ser-
vice’s target customers fall into three
categories: business professionals
looking to get quickly from point A
to B, commuters who want to beat
rush-hour traffic, and tourists.
Electric air taxis could also
serve as corporate campus shut-
tles and help shift cargo between
depots and distribution hubs, says
Roei G ­ anzarski, CEO of magniX, a
­S eattle-based firm developing
motors for electric planes. “I don’t
think we will see thousands of these
flying around each city as some
companies would like the public
to believe...but I believe eVTOLs
will play a significant part in the
future of mobility.”
However, it could take 10 to
15 years for this vision to become
reality, says Ganzarski. Among the
hurdles he cites are battery power,
regulatory issues, and the ability of
fully autonomous aircraft to han-
dle emergencies.
Pilots act as a fail-safe in many
respects, says pilot and aviation
professor emeritus Jason M ­ iddleton
from the University of New South
Wales, in Sydney. “Weather is
unpredictable; it can quickly
develop from nothing into a rag-
ing thunderstorm,” he says. “Who’s
going to predict where [air taxis]
can or can’t fly? And what happens
when they’re in the air and can’t go
to their destination?”
One answer is unmanned aircraft
system traffic management plat-
NEWS
08  | DEC 2019 | SPECTRUM.IEEE.ORG
forms, or UTMs for short, which
are already in place for drones.
­Volocopter is looking to use UTMs
to govern its air taxis. “You can take
most of the airspace-management
techniques we use in drones and
apply it to air taxis,” says Pamir
Sevincel, who leads urban air
mobility strategy at A ­ irMap, which
counts Volocopter as a client.
AirMap has developed UTM capa-
bilities for drones, such as digital
flight plans, aircraft surveillance
and monitoring, and dynamic
rerouting during emergencies. The
California-based company plans
to enable pilots or fleet managers
for drones and electric air taxis to
update flight trajectories based on
an automated assessment of risk
and potential safety issues along
planned routes.
Infrastructure—vertiports with
passenger lounges, as well as
­b attery-charging and aircraft
m a i n t e n a n c e s t a t i o n s — mu s t
also be built before electric air
taxis c an become a commer-
cial reality. V ­ olocopter has part-
nered with Skyports, a ­B ritish
infrastructure firm that recently
unveiled the first prototype of its
­VoloPort—the air taxi equivalent of
a helipad—in Singapore.
Volocopter’s Reuter says his firm
is working closely with global avia-
tion authorities, and he’s well aware
that public acceptance of autono-
mous transport will be key. “Many
people picture the skies becoming
dark and aircraft whizzing around
the city without any control or
rules. That’s a very negative and
chaotic image,” he says. “But let’s
take it step-by-step and evaluate
how it goes.” —Sandy Ong
An extended version of this article
appears in our Cars That Think blog.
↗ POST YOUR COMMENTS at https://spectrum.ieee.
org/airtaxi1219
AT LAST,
WAVE
ENERGY
TECH PLUGS
INTO
THE GRID
A buoy will feed power
to Oahu this month
Ocean waves are powerful
and perpetually replenished.
But unlike the wind and sun,
waves remain a largely untapped source
of renewable energy, despite their enor-
mous potential. A slew of projects is start-
ing to change that, with large prototypes
launching near coastlines worldwide.
In Hawaii, the OceanEnergy Buoy
is slated to connect to the island of
Oahu’s electric grid this month. The
­749-metric-ton device was recently towed
from Portland, Ore., to the U.S. Navy’s
Wave Energy Test Site, where the bright
yellow buoy will undergo a year of per-
formance tests. The project builds on a
decade of research and several smaller
iterations, including a quarter-scale
model that was tested for three years in
Ireland’s Galway Bay.
“The difficulty has been in developing
a technology that actually survives in the
marine environment, which can be very
harsh,” said John McCarthy, CEO of the
Irish buoy maker OceanEnergy.
To limit seawater effects, McCarthy’s
team designed a device that puts mechan-
ical parts above the surface. The “oscil-
lating water column” system features
a semi-submerged chamber, inside of
which an air pocket is trapped above a
 column of water. When waves crest and
water enters the chamber, it forces the air
upward, spinning a Siemens subsidiary’s
turbine system to generate electricity. As
water recedes, it creates a vacuum that
sucks in outside air and continues driv-
ing the turbine.
The 1.25-megawatt buoy will be
moored to a 60-meter-deep berth and
should withstand gale-force winds and
extreme waves. A subsea cable will link
it to Hawaiian Electric’s grid, which still
runs primarily on imported oil.
About 100 people built the buoy over
14 months at the Swan Island shipyard
in Portland, said Tom Hickman of U.S.
shipbuilder Vigor Industrial. Workers
cut, formed, and welded steel plates
into three massive sections to form the
L-shaped hull, then installed mechanical
and electrical components. On a crisp
October morning, company leaders and
dignitaries held a completion ceremony,
days before a tugboat dragged the buoy
up the Columbia River and across the
Pacific Ocean.
Tyler Gaunt, a project manager for
Vigor, said he was proud to have suc-
cessfully finished the project but happy
to see the device leave. Constructing a
first-of-its-kind prototype at a large scale
meant constantly solving problems under
a relatively tight deadline. For instance,
the supportive steel “stiffeners”
that are typically applied inside
ship hulls went on the buoy’s
exterior, to avoid creating drag
within the air chamber.
“It was essentially the oppo-
site of how we would normally
construct a ship,” he said from
the shipyard.
Globally, about 19 megawatts
of “wave energy converters”
were deployed from 2010 to
2018, though some devices
were decommissioned after
pilot tests, according to Ocean
Energy Europe (an industry
organization not connected
with OceanEnergy). The bulk
OCEANENERGY
SURF’S UP: OceanEnergy’s buoy
would be one of many such devices
on a utility-scale wave farm.
of projects have been in the United King-
dom and Western Europe, with other
devices deployed in China, Australia,
New Zealand, and the United States.
Wave energy is one of several technol-
ogies that harness the ocean’s natural
features—tides, winds, water tempera-
tures, salinity—and could provide sig-
nificant amounts of clean electricity.
Waves off U.S. shores represent some
2.64 trillion kilowatt-hours in theoreti-
cal annual energy potential—equivalent
to about two-thirds of the nation’s elec-
tricity generation in 2018, according to
an estimate by the U.S. Department of
Energy. This resource is abundant at
higher latitudes, where colder tempera-
tures and weak sunlight make it harder
to operate other renewables during cer-
tain months.
“Wind and solar are really cheap and
ubiquitous on land, but there are chal-
lenges with those technologies,” said
Bryson Robertson, codirector of the
Pacific Marine Energy Center and an
­Oregon State University associate pro-
fessor. “In places with very aggressive
decarbonization agendas, we’re going
to need all renewable resources to
really start to mitigate our impact on
the climate.”
Marine technologies still face signifi-
cant hurdles to achieving commercial
scale. It’s not yet clear how spinning tur-
bines and rotating blades will affect wild-
life. Supporting infrastructure, such as
offshore grid connections, isn’t widely
available. Licensing and permitting pro-
cesses must first ensure that devices don’t
obstruct commercial fishing, whale
watching, or other activities.
Such issues have stifled investment,
so public agencies and research insti-
tutions are leading the way, with over
a dozen testing hubs worldwide. In
­S cotland’s Orkney Islands, the Euro-
pean Marine Energy Center has
­13 ­grid-connected berths for wave and
tidal devices. New sites are under way in
Western Australia and Jeju Island, South
Korea. At the U.S. Navy hub in Hawaii,
three other developers—­C olumbia
Power Technologies, Northwest Energy
Innovations, and Oscilla Power—are also
expected to test wave energy convert-
ers starting in 2021.
Robertson said OceanEnergy’s yellow
buoy represents a valuable “data point”
in the broader effort to improve perfor-
mance and drastically reduce electric-
ity costs from marine technologies. “We
need to start putting these devices in the
water so we can start to learn lessons,”
he said. —Maria Gallucci
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/
wave-energy1219
SPECTRUM.IEEE.ORG | DEC 2019 |  09
 MAKANI DEMOS ENERGY
KITES OVER THE NORTH SEA
Airborne flyers are easier to deploy than
traditional wind turbines

The idea is simple: Send kites buoy to operate. A traditional wind tur-
or tethered drones hundreds bine must be firmly rooted to structures
of meters up in the sky to that extend down to the seafloor. Typi-
generate electricity from the persistent cally, it can operate only in depths of less
winds aloft. With such technologies, it than 50 meters, as Makani’s technical pro-
might even be possible to produce wind gram manager, Doug McLeod, said at the
energy around the clock. Airborne Wind Energy Conference 2019
Dozens of companies and researchers (AWEC 2019). McLeod explained that tra-
are developing technologies that pro- ditional offshore wind power is impossi-
duce wind power with objects adrift ble for people in coastal regions that lack
high in the sky. In October, many of shallow waters.
them gathered on the ground at a con- The buoy for the M600 test flyer was
ference in Glasgow. They presented made from existing oil and gas platform HIGH AS A KITE: The 26-meter wingspan of
experiments, field tests, and simula- materials. The flyer is an ­eight-propeller Makani’s M600 flyer is roughly three-quarters
as long as a single blade on a traditional wind
tions demonstrating various technolo- monoplane drone whose propellers loft turbine. When released from the buoy, the
gies collectively described as airborne the drone into the sky from its vertical M600 positions itself to face strong crosswinds.
wind energy (AWE). resting position on the buoy.
In August, Alameda, Calif.–based Once the flyer has achieved alti- flying thousands of flight hours in simula-
Makani Technologies flew its airborne tude—its tether extends 500 meters— tion to continue to de-risk our technology
wind turbines—which the company calls the propellers switch off and become in advance of commercial operation.”
energy kites—in the North Sea, some miniature wind turbines. Roland Representatives from the world’s
10 kilometers off the coast of Norway. Schmehl, co-organizer of AWEC 2019 next-largest AWE system—a tethered
According to Makani CEO Fort Felker, the and associate professor of aerospace flyer from Ampyx Power, in The Hague,
North Sea trials consisted of a launch and engineering at Delft University of Tech- ­Netherlands—also presented a progress
“landing” test followed by a flight test, in nology in the Netherlands, says the flyer’s report at AWEC 2019.
which the kite stayed aloft for an hour. eight 80-kW rotors make for an impres- With a 250-kW flyer called AP3, Ampyx
The Alphabet subsidiary has conducted sive system. “The sheer size of the sys- is finalizing its code and prototype. For
onshore flights of various incarnations of tem is something that is very hard to even now, the company is still testing and
its energy kites in California and Hawaii, imagine for most other startups,” he says. developing the AP3 at its headquarters.
but the North Sea flights were its first off- Makani CEO Felker says the main point Ampyx expects to perform test flights of
shore tests of a kite-and-buoy setup. The of the August test flights was to collect the AP3 in Ireland next year.
600-kilowatt kite Makani tested there data that Makani engineers can now use “We want to really experience what
flew in crosswinds—“our system’s power- to run many more simulated test flights it takes to install a commercial-size air-
generating mode,” Felker says. “Our test to further develop their system. borne wind-energy system in our AP3
site in Hawaii is focused on maturing the “The successful flights validated that project,” said Sören Sieberling, project
energy kite system for continuous hands- our models for launch, landing, and manager at Ampyx.
MAKANI TECHNOLOGIES

off operation,” he adds. crosswind flight from a floating plat- Unlike the Makani M600, which gen-
Makani’s 26-meter M600 prototype, cre- form were indeed accurate,” he says. erates wind energy via spinning rotors
ated with support from Royal Dutch Shell “This means we can confidently use our at altitude—and then sends electricity
and others, requires only an anchored modeling tools to test system changes— down through the tether—the Ampyx AP3

NEWS

10  | DEC 2019 | SPECTRUM.IEEE.ORG

uses the wind’s power to tug on the line. So the AP3’s
electric generator, which is driven by the tether’s
motion, stays on the ground. The kite must there-
fore spend some of its flight time reeling the line in
so it can tug the line back out again.
The AWE field, says conference co-organizer
Schmehl, is today active with many research pro-
grams and startups. Schmehl is an advisor to a com-
pany based in the Netherlands called ­Kitepower,
which is developing a 100‑kW AWE system. A
­German company, Kiteswarms, is attaching multi-
ple drones to the end of a tether, hoping that strat-
egy will add capacity and improve reliability as it
scales up its own model. And Norway’s Kitemill
and Switzerland’s Twingtec use vertical takeoffs
and landings to loft their flyers to altitudes where
they can begin generating electricity.
Ultimately, the list of AWE projects will nar-
row, Schmehl predicts, down to technologies that
solve real-world problems. “We as an industry now
have to start earning money—and stop develop-
ing for the ultimate, perfect end product,” he says.
—M a r k A nderson
An extended version of this article appears in our
Energy­wise blog.
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/energykites1219
JOURNAL WATCH
It’s Personal: Bioacoustic
Identification
Every sound has a unique signature. Researchers
in South Korea are exploring whether it’s possible
to recognize people from the sound waves that
pass through their bodies. The team’s new study,
published in IEEE Transactions on Cybernetics,
suggests bioacoustic signatures can accurately
identify a person 97 percent of the time.
The biometric system developed by the group at
the Electronics and Telecommunications Research
Institute (ETRI) uses a transducer to generate
vibrations, and aims the resulting sound waves
at a subject’s finger. After the sound has passed
through, a sensor captures that individual’s unique
bioacoustic signature, which is further characterized
through modeling.
“Modeling allowed us to infer what structures
or material features of the human body actually
differentiated people,” explains Joo Yong Sim, one of the
ETRI researchers. “For example, we could see how the
structure, size, and weight of the bones, as well as the
stiffness of the joints, affect the bioacoustic spectrum.”
The approach is effective enough to recognize
different fingers on the same hand. The researchers
were initially concerned that the system’s accuracy
would diminish over time due to changes in the
body. To test this, they acquired acoustic data from
participants on three occasions, spaced 30 days apart.
“We were very surprised that people’s bioacoustic
spectral pattern maintained well over time,” says
Sim. “These results suggest that the bioacoustic
signature reflects more anatomical features than
changes in water, body temperature, or biomolecule
concentration in blood that change from day to day.”
The accuracy of this method does not yet match
those of fingerprint or iris scans. Sim’s team plans
to add sensors to boost the system’s accuracy and
hopes to commercialize the technology in phones and
wearables. And in an unexpected twist, the technique
proved so accurate at analyzing tissues that the
team is now exploring whether it could diagnose
musculoskeletal diseases. —MICHELLE HAMPSON
An extended version of this article appears on our
website in the Journal Watch section.
SPECTRUM.IEEE.ORG | DEC 2019 |  11
4 PRODUCTS TO
More data plus the need for fewer
repeaters under the ocean would jus-
tify a higher price, says Austin Jordan of

MANUFACTURE IN ORBIT Made in Space, which plans to produce

Organs and fiber-optic cables will justify
the “made in space” label
Space is open for business,
and some entrepreneurs
plan to make the final fron-
tier into a manufacturing hub. There’s
plenty of real estate. But it takes a few
thousand dollars to launch a kilogram
of stuff into space.
“The key question is: What is it that jus-
tifies the expense of doing these things in
low Earth orbit?” says William Wagner,
director of the University of Pittsburgh’s
McGowan Institute for Regenerative
Medicine, which will conduct biomedi-
cal research on the International Space
­Station (ISS).
Here are some technologies that might
merit the “made in space” label.
FIBER-OPTIC CABLE Made from fluo-
ride glass, a kind of fiber-optic cable called
ZBLAN could have as little as one-tenth
the signal loss of silica-based optical fibers.
But quality ZBLAN fibers are hard to
make on Earth. As the molten glass is
stretched into fibers as thin as fishing line
and then cooled, tiny crystals sometimes
form, which can weaken signals. Micro-
gravity suppresses the formation of these
crystals, so fibers made in space would
carry more data over longer distances.
such fibers in space for terrestrial clients.
“The math works. It would pay for itself
and drive a profit,” he says.
Two other companies, Fiber Optic Man-
ufacturing in Space and Physical Optics
Corp., also plan to make ZBLAN fibers
in low Earth orbit.
O R GA N S There are 120,000 people
waiting for an organ transplant in the
United States alone. “Most will never see
one, there is such a shortage,” says Eugene
Boland, chief scientist at ­Techshot, which
wants to print human hearts in space.
The heart, with its four empty cham-
bers and highly organized muscle tissue
made of different types of cells, is virtu-
ally impossible to print on the ground.
On Earth, tissues printed with runny
bioinks made of gel and human stem
cells collapse under their own weight.

ROBOTIC END - EFFECTORS

Measure all six components of force

and torque in a compact, rugged sensor.

Interface Structure
high-strength alloy provides IP60, High-Speed Electronics
IP65, and IP68 environmental interfaces for Ethernet, PROFINET,
protection as needed EtherNet/IP, Analog, USB, CAN
EtherCAT, Wireless, and more
Sensing Beams and Flexures
designed for high stiffness and over- Silicon Strain Gages
load protection without compromising provide high noise immunity,
resolution accuracy, and high factor-of-safety,
standard on all F/T models

Engineered for high-performance and maximum stiffness,

with the highest resolution and accuracy available, it’s the
ultimate force/torque sensor. Only from ATI.

www.ati-ia.com
919.772.0115

12  | DEC 2019 | SPECTRUM.IEEE.ORG
Scientists must add toxic chemicals or
a scaffold.
Printing hearts and other organs in
microgravity could be done using pure
bioinks. “The cylindrical shape extruded
from the nozzle is maintained, so you
can build a more fragile 3D structure that
would allow cells in the gels to secrete
their own matrix and strengthen up,”
says Wagner. And the printed layers fuse
tech startup Aleph Farms plans to do with
meat. The two-year-old Rehovot-based
company grows cultured beefsteaks that
look and taste like the real thing. “While
other companies use only muscle cell,
we also grow connective tissue, blood
vessels, and fat cells, which lets us make
beefsteaks instead of patties,” says Yoav
Reisler, external relations manager at
the company.
In September, the company teamed
up with Russian company 3D Bioprint-
ing Solutions to create the first tiny piece
of meat on the ISS. It isn’t a huge techni-
cal advance, but it could feed astronauts
on long-term crewed missions, as well as
future space settlers as they set up a per-
manent base. —Prachi Patel
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/
madeinspace1219

together without forming the striations

seen in constructs printed on the ground,
Boland says.
Techshot, which is based in Greenville,
Ind., is partnering with 3D-bioprinter
manufacturer nScrypt. Their first bio-
printer went to the ISS in July, but the
small patch of heart muscle it printed
didn’t survive reentry. The next mission,
which launched in November, should
result in thicker tissue that can be tested
on Earth when it returns in January.

METAL ALLOYS Outer space is the per-

fect place to make metal alloys. Micro-
gravity allows the metals and other
elements to mix together more evenly.
Magnesium alloys for medical implants
have especially high potential. At half the
weight of titanium alloys, magnesium
alloys more closely match the density and
strength of bone, and they harmlessly
biodegrade in the body, says University
of Pittsburgh bioengineering professor
Prashant Kumta, who is working with
Techshot to produce his patented alloys
in a high-temperature furnace on the ISS.
Making these alloys involves melting
highly reactive magnesium with other
elements such as calcium and zinc, keep-
ing the melted materials in a vacuum for Over 50 New Features & Apps in this New Version!
a long time so the elements mix evenly,
and then cooling it all down. Over 500,000 registered users worldwide in:
On Earth, impurities settle to the bot-
■ 6,000+ Companies including 20+ Fortune Global 500
tom, and the upper layer oxidizes to form
■ 6,500+ Colleges & Universities
an unusable skin. Both have to be thrown
out. Even the usable middle layer has
■ 3,000+ Government Agencies & Research Labs
pores and pockets of unmixed elements
®

and must be further processed to make

a quality material. None of these prob-
www.originlab.com
lems occur when alloys are manufactured
in microgravity.
25+ years serving the scientific and engineering community.
MEAT What Techshot and nScrypt want For a 60-day FREE TRIAL, go to OriginLab.Com/demo and enter code: 8547
to do with human organs, Israeli food-

SPECTRUM.IEEE.ORG | DEC 2019 |  13
2019 HOLIDAY
GIFT GUIDE
TECHIE GIFTS FOR
ALL BUDGETS

S IEEE SPECTRUM REPORTED AT

A the end of last year, pinball is hav-
ing a revival, driven in part by the
shift to e-commerce, which is turning erst-
while big-box retail stores into cheap real es-
tate for family entertainment centers. Modern
pinball machines, with enhancements like up-
gradable software, are vastly more sophisti-
cated than their electromechanical ancestors.
Stern Pinball is in the vanguard of this renais-
sance, making home and arcade versions of
many of its games. The latest title available in
a home version is the US $4,500 Star Wars Pin,
based on comics artwork and models inspired
by the original movie trilogy. —STEPHEN CASS

STERN PINBALL (2)

14  | DEC 2019 | SPECTRUM.IEEE.ORG
 Wichit Sirichote is a professor at King Mongkut’s
2 Institute of Technology, in Bangkok. He’s also the
maker of a terrific line of single-board computers based
on classic CPUs such as the 6502, Z80, and 8088.
­Prices range from $85 to $175, depending on the CPU.
They are bare-bones machines designed for learning and
proto­typing, but they are very flexible: You can upload
code through an RS-232 port, plug in a standard LCD
character display directly using an onboard connector,
and add other custom hardware via a bus-expansion slot.
Sirichote wrote his own monitor software for the boards
that lets you, for example, examine the contents of CPU
registers, and extensive documentation is available.

Artie 3000 is a modern version of the classic turtle drawing robot that could
3 be found in classrooms connected to a microcomputer running the Logo
language. But what’s really nice about the $70 Artie is that unlike with the turtles
of old, you can easily program it in a variety of ways. At the most basic level, you can
control Artie and its pen directly, remote-control style, and then move on up through
various graphical block languages and into full-fledged Python and JavaScript.
Budding programmers benefit from a gentle learning curve, and advanced coders
interested in procedurally generated art get access to a cheap robot.
FROM TOP: RANDI KLETT; EDUCATIONAL INSIGHTS; ONETESLA (2)

It’s not going to win awards for the

4 quality of its sound, but it is a crowd
pleaser. OneTesla’s $400 Musical Tesla Coil
Kit can be driven directly by a MIDI-enabled
instrument or play a MIDI file. The frequency
of the notes is used to modulate the output
of the coil with a square wave, producing
buzzing notes and impressive sparks over
half a meter long. A smaller version is also
available for $230.

POST YOUR COMMENTS at https://spectrum.ieee.org/giftguide1219 SPECTRUM.IEEE.ORG | DEC 2019 |  15

 RESOURCES_GIFT GUIDE

Doppel is worn like a watch, except you wear it on the inside of

5 your wrist rather than on the outside. A rotating weight inside cre-
ates a rhythmic vibration. The purpose of the $280 wearable is to im-
prove focus and reduce stress, by using rhythms that are faster or slower,
respectively, than your normal resting heart rate. (The accompanying
smartphone app measures your heart rate when you place your finger
over your phone’s camera lens and looks at changes in the ambient light
that gets filtered through.) It did help reduce my anxiety levels some-
what, but I found it worked best with deliberate mindfulness techniques,
so if you’re not already familiar with those, your mileage may vary.

Minecraft is already used to introduce

6 children to writing software. The
$300 Piper Computer Kit, aimed at 8- to
13-year-olds, extends that idea to hardware.
Kids first assemble the wooden case and
plug together the basic components of the
kit, which is based on a Raspberry Pi and
comes with its own screen. While the kit in-
cludes a mouse, there is no keyboard. In-
stead a breadboarding module is provided,
which can be used to, for example, wire up
buttons to control events in Minecraft
through a series of game levels.

If you’re looking for a nonfiction book to give, try one from the
7 Platform Studies series from MIT Press. These books describe
influential platforms in the history of digital media, examining how the
specific technical details and hardware capabilities of each platform
(or, in academia-speak, “affordances”) shaped the software that ran on
them and how that combination in turn affected the industry and wider
culture. The most recent 2019 title (The Media Snatcher) dissects the
PC Engine/TurboGrafx-16 console. The highlights of the series so far
for me are Racing the Beam, about the Atari 2600, and Minitel.
In the fiction department, Spectrum’s recommendation is Fall, or
Dodge in Hell (William Morrow, 2019) by Neal Stephenson, author of
the cyberpunk classic Snow Crash. Fall is a sequel of sorts to his 2011
novel, Reamde, but it can be read completely independently of Reamde
(and is in fact a much better book). Reamde is an entertaining enough
techno­thriller, but Fall is Stephenson at his best, weaving together deep
philosophical questions against the background of a compelling vision
of the future (a chapter featuring a journey across an America that’s
been utterly fragmented by competing social-media feeds is plausibly
chilling). In Fall, the lead character awakens in a digital afterlife, in which
his first order of business is to create a universe to live in.
CLOCKWISE FROM TOP: DOPPEL; PIPER; HARPERCOLLINS; MIT PRESS (3)

16  | DEC 2019 | SPECTRUM.IEEE.ORG
 RESOURCES_HANDS ON

HACK YOUR OLD POINT-AND-SHOOT RESURRECTED TECH:

With a Raspberry Pi and the

ADD FEATURES TO CANON CAMERAS right software, an old digital

camera can be network

WITH NEW FIRMWARE enabled and used for

projects unsuitable for either
a smartphone camera or an
expensive DSLR.

script that allows a more fine-grained incre-

DECADE IS A LONG TIME IN TECHNOLOGY—LONG ENOUGH
A
ment of 1 second, but I wanted to try time-
for a technology to go from hot product to conspicuously ob-
lapse photography of the Empire State
solete to retro cool. In 2010, IEEE Spectrum’s David Schneider
Building, which we happen to have a good
wrote about a hack to supplant the firmware in Canon point-and-shoot
view of from Spectrum’s New York office.
cameras and add new features, such as motion detection. As it turns out,
During the day, the light changes slowly, so
at the time point-and-shoot cameras were near their zenith of popular-
I wanted to shoot one photo every few min-
ity. Since then, while compact stand-alone digital cameras are still being
utes. At dusk, however, the lighting on and
made, their sales have shrunk dramatically. As the smartphone camera
around the building changes more dramati-
became the most ubiquitous type of camera on the planet, point-and-
cally, so I wanted photos taken at a faster rate.
shoot cameras found themselves relegated to the back of the closet.
The first thing was to test my camera. It’s a
credit to Canon that despite years of disuse,
all the parts sprang to life. The only problem
That was certainly the case with our Canon a task for which even an 11-year-old digital was on the battery side. I had three batteries,
PowerShot S80. My wife bought it in 2008 camera, with its larger optics, can compete one of which refused to charge at all, and
primarily to document her paintings in be- with today’s smartphones on image quality. two others I no longer trusted for a long­-
tween professional photo shoots, and a few This scenario makes mobility a moot point, duration experiment, so I found a DC adapt-
years later we replaced it with a mirrorless but the task also requires more sophistica- er on eBay that powers the camera from a
Nikon 1 J-1 with interchangeable lenses. So tion than even CHDK—the open-source firm- wall socket.
when I found the S80 while decluttering re- ware replacement David Schneider wrote Then I installed CHDK. Fortunately, this is
cently, I wondered: Was it just e-waste now, about in 2010—can easily offer alone. one of those rare pieces of open-source soft-
or could it be combined with today’s technol- My S80’s original firmware had a func- ware for which the documentation is a com-
ogy to do interesting things? tion that would take a photograph at fixed prehensive and intelligible wiki. Looking up
RANDI KLETT

I decided the perfect test case for my S80 intervals of between 1 and 60 minutes, in the instructions for my S80, I determined
was variable time-lapse photography. This is 1­ -minute increments. CHDK provides a its current firmware, which turned out to be

SPECTRUM.IEEE.ORG | DEC 2019 |  17
 RESOURCES_HANDS ON

1.00f. Only the 1.00g version is compatible
with CHDK, so I followed the instructions to
upgrade the factory firmware, the biggest
obstacle to which was finding the right util-
ity software to open the 7z format that the
firmware file was compressed with.
A cross-platform tool called Stick makes
finishing the CHDK install easy: Drop a photo
taken with a camera onto the tool’s interface
and it analyzes the metadata and downloads
the exact version of CHDK required onto a
CHDK provides an interface for remote EMPIRE STATES: A program triggered the S80 at
control of the camera via the USB link normal- different rates to capture slow changes during the
day and more rapid changes at dusk. CHDK adds
ly used to download photographs directly to many other features [far right] to old cameras.
a PC. A number of programs can use this PTP
standard, including chdkptp, which offers both The first was that chdkptp couldn’t connect
a command line version and a graphical user to the S80—a helper process on the Pi was
interface (GUI) version that lets you see what grabbing the connection, assuming I want-
is being displayed in the viewfinder screen live. ed to download photos. The simplest solu-
One of the nice things about chdkptp is that tion was to find the offending process using
a precompiled binary, bundled with required a “ps -ax | grep gphoto2” command, and
supporting libraries, is available for the Rasp- “kill -9” it. (This works only on a per-session

STEPHEN CASS (4)

SD card. Launching CHDK on my camera berry Pi, thus eliminating dependency hell. basis; if you want to permanently disable the
just requires putting the prepared card in and I ran into two problems, which were re- helper, you’ll have to edit some deep con-
pressing the S80’s “shortcut” button. solved after searches of CHDK’s user forums. figuration files.)

AGM2222 Series
High-efficiency Power Inductors
• Ideal for bi-directional 12 V – 48 V DC-DC converters
• Isat ratings up to 110 Amps with very low DCR
• Package measures just 22 x 22 x 22 mm
• 13 inductance values from 1.9 to 10 µH

Free Samples @ www.coilcraft.com

18  | DEC 2019 | SPECTRUM.IEEE.ORG
 Spectrum-GC19-2-3-Page-Ad_v2_12NOV19_JB_Print-Ready.pdf 2 11/14/19 10:44 AM

IEEE Global Communications Conference

9-13 December 2019 // Waikoloa, Hawaii, USA
Revolutionizing Communications

JOIN IEEE GLOBECOM 2019

IN HAWAII!

My camera and chdkptp could now con-

nect, but I still couldn’t actually take a pho-C

to. This was solved by writing a script with

M

some mode commands I found on a forum.

Y
CHDK runs the script when I press the cam-
era’s shutter, and then it is happy to accept CM

remote commands. MY

To implement my variable time-lapse CY

schedule, I wrote a short Python program on CMY

the Pi. I looked up the time of the sunset and

K
Expert presentations on latest in
set the Python program to check the clock.
Outside a half-hour window around sunset 5G, AI, ML, Wireless Communications
it would take a photo every 10 minutes, and
one every 30 seconds inside the window. To technologies and more.
control the S80, I just issued an OS call to
the chdkptp command line tool that con-
nected and triggered the shutter—that is,
os.system(“./chdkptp.sh -c -eshoot”).
I left the system running from the early af-
ternoon till dusk, and when I returned I had
113 images, which I dumped into iMovie to
make a time-lapse video. Ta-da!
Now that I have my proof of concept, it Learn more and register today at:
would be a straightforward task to write
a Python script that could download the https://globecom2019.ieee-globecom.org/
times of sunrise and sunset and adjust itself
automatically. I can also save images direct- SPONSORED BY:
ly to the Pi. Then I could access and down-
load these images remotely over a wireless IEEE Communications Society
network, allowing for the option of leaving
the camera and Pi in place for long peri- www.ComSoc.org
ods of time for truly epic time-lapse movies.
—STEPHEN CASS

POST YOUR COMMENTS at https://spectrum.ieee.org/

canonhack1219

SPECTRUM.IEEE.ORG | DEC 2019 |  19
 INTERNET OF EVERYTHING_BY STACEY HIGGINBOTHAM OPINION

Microsoft has some ideas about what

the documentation process should look
like. The process starts with the researcher
structuring and organizing the raw data
and annotating it appropriately. Not hav-
ing a documented process at this stage
could lead to poorly annotated data that
has biases associated with it or is unrelated
to the problem the business wants to solve.
Next, during training, a researcher feeds
the data to a neural network and tweaks
how it weighs various factors to get the
desired result. Typically, the researcher is
still working alone at this point, but other
people should get involved to see how the
model is being developed—just in case
questions come up later during a compli-
ance review or even a lawsuit.
A neural network is a black box when
it comes to understanding how it makes
its decisions, but the data, the number
of layers, and how the network weights
different parameters shouldn’t be mys-
terious. The researchers should be able
to tell how the data was structured and

SHOW YOUR MACHINE- weighted at a glance.

It’s also at this point where having good
documentation can help make a model

LEARNING WORK more flexible for future use. For example,

a shopping site’s model that crunched
data specifically for Christmas spending
IN THE LAST TWO YEARS, the U.S. Food and Drug Administration has patterns can’t apply that same model to
approved several machine-learning models to accomplish tasks such as Valentine’s Day spending. Without good
classifying skin cancer and detecting pulmonary embolisms. But for the documentation, a data scientist would
companies who built those models, what happens if the data scientist who have to essentially rebuild the model,
•
wrote the algorithms leaves the organization? In many businesses, an individual or rather than going back and tweaking a few
a small group of data scientists is responsible for building essential machine-learning parameters to adjust it for a new holiday.
models. Historically, they have developed these models on their own laptops through The last step in the process is actu-
trial and error, and pass it along for production when it works. But in that transfer, ally deploying the model. Historically,
the data scientist might not think to pass along all the information about the model’s only at this point would other people
•
development. And if the data scientist leaves, that information is lost for good. That get involved and acquaint themselves
potential loss of information is why experts in data science are calling for machine with the data scientist’s hard work. With-
learning to become a formal, documented process overseen by more people inside out good documentation, they’re sure
•
an organization. Companies need to think about what could happen if their data to get headaches trying to make sense
scientists take new jobs, or if a government organization or an important customer of it. But now that data is so essential
asks to see an audit of the algorithm to ensure it is fair and accurate. Not knowing to so many businesses—not to mention
what data was used to train the model and how the data was weighted could lead to the need to adapt quickly—it’s time for
a loss of business, bad press, and perhaps regulatory scrutiny, if the model turns out companies to build machine-learning
•
to be biased. David Aronchick, the head of open-source machine-learning strategy processes that rival the quality of their
at Microsoft Azure, says companies are realizing that they must run their machine- ­software-development processes. n
learning efforts the same way they run their software-development practices. That
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/
means encouraging documentation and codevelopment as much as possible. machinelearningwork1219

20  | DEC 2019 | SPECTRUM.IEEE.ORG ILLUSTRATION BY Greg Mably

 NUMBERS DON’T LIE_BY VACLAV SMIL OPINION

Gas turbines are now much more pow-

erful. Siemens now offers a CCGT for util-
ity generation rated at 593 MW, nearly
40 times as powerful as the Neuchâtel
machine and operating at 63 percent
efficiency. GE’s 9HA delivers 571 MW in
­s imple-cycle generation and 661 MW
(63.5 percent efficiency) by CCGT.
Their near-instant availability makes gas
turbines the ideal suppliers of peak power
and the best backups for new intermittent
wind and solar generation. In the United
States they are now by far the most afford-
able choice for new generating capacities.
The levelized cost of electricity—a measure
of the lifetime cost of an energy project—
for new generation entering service in
2023 is forecast to be about US $60 per
megawatt-hour for coal-fired steam turbo
generators with partial carbon capture,
$48/MWh for solar photovoltaics, and
$40/MWh for onshore wind—but less than
$30/MWh for conventional gas turbines
and less than $10/MWh for CCGTs.
Gas turbines are also used for the com-

SUPEREFFICIENT bined production of electricity and heat,

which is required in many industries and
is used to energize central heating systems

GAS TURBINES in many large European cities. These tur-

bines have even been used to heat and
light extensive Dutch greenhouses, which
EIGHTY YEARS AGO, the world’s first industrial gas turbine began additionally benefit from their use of the
to generate electricity in a municipal power station in Neuchâtel, generated carbon dioxide to speed up
­Switzerland. The machine, installed by Brown Boveri, vented the exhaust the growth of vegetables. Gas turbines
without making use of its heat, and the turbine’s compressor consumed also run compressors in many industrial
nearly three-quarters of the generated power. That resulted in an efficiency of just enterprises and in the pumping stations
•
17 percent, or about 4 MW. The disruption of World War II and the economic of long-distance pipelines. The verdict
difficulties that followed made the Neuchâtel turbine a pioneering exception until is clear: No other combustion machines
1949, when Westinghouse and General Electric introduced their first low-capacity combine so many advantages as do mod-
designs. There was no rush to install them, as the generation market was dominated ern gas turbines. They’re compact, easy
by large coal-fired plants. By 1960 the most powerful gas turbine reached 20 MW, to transport and install, relatively silent,
still an order of magnitude smaller than the output of most steam turbo generators. affordable, and efficient, offering nearly
• In November 1965, the great power blackout in the U.S. Northeast changed many instant-on power and able to operate with-
minds: Gas turbines could operate at full load within minutes. But rising oil and gas out water cooling. All this makes them the
prices and a slowing demand for electricity prevented any rapid expansion of the unrivaled stationary prime mover.
•
new technology. The shift came only during the late 1980s. By 1990 almost half And their longevity? The Neuchâtel
of all new installed U.S. capacity was in gas turbines of increasing power, reliabil- turbine was decommissioned in 2002,
ity, and efficiency. But even efficiencies in excess of 40 percent—­matching today’s after 63 years of operation—not due to
best steam turbo generators—produce exhaust gases of about 600 °C, hot enough any failure in the machine but because
to generate steam in an attached steam turbine. These combined-cycle gas turbines of a damaged generator. n
(CCGTs) arrived during the late 1960s, and their best efficiencies now top 60 per-
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/
cent. No other prime mover is less wasteful. gasturbine1219

ILLUSTRATION BY Stuart Bradford

SPECTRUM.IEEE.ORG | DEC 2019 |  21
Augmented Reality for Robocars
Adding simulated vehicles to closed-course testing 23
makes robocars safer | BY YI H E NG F EN G A N D HE N RY X. L I U
DEC 2019
SPECTRUM.IEEE.ORG

SIMCITY FOR CARS: To test connected self-driving cars you need more than just a
paved track—you need an entire complement of city scenes, including storefronts
and steel manhole covers (to test the discrimination of the car’s radar). Mcity, on
the North Campus of the University of Michigan, in Ann Arbor, goes further by
wirelessly connecting a self-driving, sensor-laden Lincoln MKZ test car to a road
infrastructure that itself can simulate vehicles. The resulting augmented reality
enables the test car to interact with these “ghost cars.”
FOR DECADES, A N YO N E W H O WA NT E D TO K NOW W H E T H E R A N E W CAR WAS SAF E TO D R I V E C O U LD SI M P LY P U T I T T H R O U G H I TS PACES,
using tests established through trial and error. Such tests might investigate whether the car can take a sharp
turn while keeping all four wheels on the road, brake to a stop over a short distance, or survive a collision
with a wall while protecting its occupants. • But as cars take an ever greater part in driving themselves, such
straightforward testing will no longer suffice. We will need to know whether the vehicle has enough i­ ntelligence
to handle the same kind of driving conditions that humans have always had to manage. To do that, a­ utomotive
safety-assurance testing has to become less like an obstacle course and more like an IQ test.

One obvious way to test the brains as well as the brawn
of autonomous vehicles would be to put them on the road
along with other traffic. This is necessary if only because the
self-driving cars will have to share the road with the human-
driven ones for many years to come. But road testing brings
two concerns. First, the safety of all concerned can’t be guar-
anteed during the early stages of deployment; self-driving test
cars have already been involved in fatal accidents. Second is
the sheer scale that such direct testing would require.
That’s because most of the time, test vehicles will be
driven under typical conditions, and everything will go as
it normally does. Only in a tiny fraction of cases will things
take a different turn. We call these edge cases, because
they concern events that are at the edge of normal experi-
ence. Example: A truck loses a tire, which hops the median
and careens into your lane, right in front of your car. Such
edge cases typically involve a concurrence of failures that
are hard to conceive of and are still harder to test for. This
raises the question, How long must we road test a self-driving,
connected vehicle before we can fairly claim that it is safe?
The answer to that question may never be truly known.
What’s clear, though, is that we need other strategies to
gauge the safety of self-driving cars. And the one we
describe here—a mixture of physical vehicles and computer
simulation—might prove to be the most effective way there
is to evaluate self-driving cars.
A FATAL CRASH OCCURS ONLY ONCE in about 160 million kilo­meters of
driving, according to statistics compiled by the U.S. National
Highway Traffic Safety Administration. That’s a bit more than
the distance from Earth to the sun (and 10 times as much as
has been logged by the fleet of Google sibling Waymo, the
company that has the most experience with self-driving cars).
To travel that far, an autonomous car driving at highway
speeds for 24 hours a day would need almost 200 years. It
would take even longer to cover that distance on side streets,
passing through intersections and maneuvering around
parking lots. It might take a fleet of 500 cars 10 years to fin-
ish the job, and then you’d have to do it all over again for
each new design.

ALL PHOTOS: UNIVERSITY OF MICHIGAN

24  | DEC 2019 | SPECTRUM.IEEE.ORG
 Clearly, the industry must augment road testing with other
strategies to bring out as many edge cases as possible. One
method now in use is to test self-driving vehicles in closed
test facilities where known edge cases can be staged again
and again. Take, as an example, the difficulties posed by cars
that run a red light at high speed. An intersection can be built
as if it were a movie set, and self-driving cars can be given the
task of crossing when the light turns green while at the same
time avoiding vehicles that illegally cross in front of them.
While this approach is helpful, it also has limitations.
Multiple vehicles are typically needed to simulate edge cases,
and professional drivers may well have to pilot them. All
this can be costly and difficult to coordinate. More impor-
tant, no one can guarantee that the autonomous vehicle
will work as desired, particularly during the early stages of
such testing. If something goes wrong, a real crash could
happen and damage the self-driving vehicle or even hurt
people in other vehicles. Finally, no matter how ingenious
the set designers may be, they cannot be expected to create
a completely realistic model of the traffic environment. In
real life, a tree’s shadow can confuse an autonomous car’s
sensors and a radar reflection off a manhole cover can make
the radar see a truck where none is present.
Computer simulation provides a way around the limita-
tions of physical testing. Algorithms generate virtual vehicles
and then move them around on a digital map that corre-
sponds to a real-world road. If the data thus generated is
then broadcast to an actual vehicle driving itself on the same
road, the vehicle will interpret the data exactly as if it had
come from its own sensors. Think of it as augmented real-
ity tuned for use by a robot.
Although the physical test car is driving on empty roads, it
“thinks” that it is surrounded by other vehicles. Meanwhile, it
sends information that it is gathering—both from augmented
reality and from its sensing of the real-world surroundings—
back to the simulation platform. Real vehicles, simulated
vehicles, and perhaps other simulated objects, such as
pedestrians, can thus interact. In this way, a wide variety
of scenarios can be tested in a safe and cost-effective way.

THE IDEA FOR AUTOMOTIVE AUGMENTED REALITY came to us by the back

door: Engineers had already improved certain kinds of com-
puter simulations by including real machines in them. As far
back as 1999, Ford Motor Co. used measurements of an actual
revving engine to supply data for a computer simulation of
a power train. This hybrid simulation method was called
hardware-in-the-loop, and engineers resorted to it because

THE HOGWARTS EXPRESS: A central computer system [opposite] receives information from wireless stations [top left], which collect the data
from traffic signals and from real cars on the test track [top right]. The computer then uses the data to simulate cars whose movements are
broadcast back to the physical test car by way of the same wireless stations. The virtual cars can simulate cross traffic [bottom left] and such
unusual events as the running of a red light. Even an entire train can be simulated: An observer in the self-driving car will notice that the car has
come to a halt at a train crossing even though no actual train is there to be seen. The point of such simulations is to let a connected, self-driving
car accumulate experience in “edge cases”—traffic events that are as critical to survival as they are rare in the real world.

SPECTRUM.IEEE.ORG | DEC 2019 |  25
 FESTOONED DRIVE-BY- CAMERA
WITH SENSORS WIRE KIT SENSORS
The Mcity test car
carries lidar (a laser
range finder) both
for a theater-in-the- 2 LIDARS
IN FRONT,
round view (provided 1 IN REAR
by the rotating roof
tower) and for a
look‑ahead focus.
Radar complements
this sense, and GPS—
in a highly accurate
radio-corrected
form—rounds out the RADARS AT ALL 4 CORNERS,
1 RADAR IN FRONT
armamentarium.
mimicking an engine in software can be very difficult. Know-
ing this history, it occurred to us that it would be possible
to do the opposite—generate simulated vehicles as part of
a virtual environment for testing actual cars.
In June 2017, we implemented an augmented-reality envi-
ronment in Mcity, the world’s first full-scale test bed for
autonomous vehicles. It occupies 32 acres on the North
Campus of the University of Michigan, in Ann Arbor. Its
8 lane-kilometers (5 lane-miles) of roadway are arranged
in sections having the attributes of a highway, a multilane
arterial road, or an intersection.
Here’s how it works. The autonomous test car is equipped
with an onboard device that can broadcast vehicle status,
such as location, speed, acceleration, and heading, doing
so every tenth of a second. It does this wirelessly, using
dedicated short-range communications (DSRC), a standard
similar to Wi-Fi that has been earmarked for mobile users.
Roadside devices distributed around the testing facility
receive this information and forward it to a traffic-simulation
model, one that can simulate the testing facility by boiling
it down to an equivalent network geometry that incorpo-
rates the actions of traffic signals. Once the computer model
receives the test car’s information, it creates a virtual twin
of that car. Then it updates the virtual car’s movements
based on the movements of the real test car.
Feeding data from the real test vehicle into the computer
simulation constitutes only half of the loop. We complete
the other half by sending information about the various
vehicles the computer has simulated to the test car. This is
the essence of the augmented-reality environment. Every
simulated vehicle also generates vehicle-status messages at
26  | DEC 2019 | SPECTRUM.IEEE.ORG
32-CHANNEL, ONBOARD
360-DEGREE LIDAR WIRELESS UNIT
COMPUTERS
RADIO-
RADIO MODEM RADIO-CORRECTED CORRECTED
GPS UNIT GPS UNIT
a frequency of 10 hertz, which we forward to the roadside
devices, which in turn broadcast it in real time. When the
real test car receives that data, its vehicle-control system
uses it to “see” all the virtual vehicles. To the car, these
simulated entities are indistinguishable from the real thing.
By having vehicles pass messages through the roadside
devices—that is, by substituting “vehicle-to-infrastructure”
connections for direct “vehicle-to-vehicle” links—real vehi-
cles and virtual vehicles can sense one another and interact
accordingly. In the same fashion, traffic-signal status is also
synchronized between the real and the simulated worlds.
That way, real and virtual vehicles can each “look” at a given
light and see whether it is green or red.
The status messages passed between real and simulated
worlds include, of course, vehicle positions. This allows
actual vehicles to be mapped onto the simulated road
network, and simulated vehicles to be mapped into the
actual road network. The positions of actual vehicles are
represented with GPS coordinates—latitude, longitude, and
elevation—and those of simulated vehicles with local coor-
dinates—x, y, and z. An algorithm transforms one system of
coordinates into the other.
But that mathematical transformation isn’t all that’s
needed. There are small GPS and map errors, and they
sometimes prevent a GPS position, forwarded from the
actual test car and translated to the local system of coordi-
nates, from appearing on a simulated road. We correct these
errors with a separate mapping algorithm. Also, when the
test car stops, we must lock it in place in the simulation, so
that fluctuations in its GPS coordinates do not cause it to
drift out of position in the simulation.
ILLUSTRATION BY Chris Philpot
 Everything here depends on wireless communication.
To ensure that it was reliable, we installed four roadside
radios in Mcity, enough to cover the entire testing facil-
ity. The DSRC wireless standard, which operates in the
5.9­-gigahertz band, gives us high data-transmission rates
and very low latency. These are critical to safety at high
speeds and during stop-on-a-dime maneuvers. DSRC is in
wide use in Japan and Europe; it hasn’t yet gained much
traction in the United States, although Cadillac is now equip-
ping some of its cars with DSRC devices.
Whether DSRC will be the way cars communicate with
one another is uncertain, though. Some people have argued
that cellular communications, particularly in the coming
5G implementation, might offer equally low latency with a
greater range. Whichever standard wins out, the communica-
tions protocols used in our system can easily be adapted to it.
We expect that the software framework we used to build
our system will also endure, at least for a few years. We
constructed our simulation with PTV Vissim, a commer-
cial package developed in Germany to model traffic flow
“microscopically,” that is, by simulating the behavior of each
individual vehicle.
One thing that can be expected to change is the test vehi-
cle, as other companies begin to use our system to put their
own autonomous vehicles through their paces. For now,
our one test vehicle is a Lincoln MKZ Hybrid, which is
equipped with DSRC and thus fully connected. Drive-by-
wire controls that we added to the car allow software to
command the steering wheel, throttle, brake, and transmis-
sion. The car also carries multiple radars, lidars, cameras,
and a GPS receiver with real-time kinematic positioning,
which improves resolution by referring to a signal from a
ground-based radio station.
WE HAVE IMPLEMENTED TWO TESTING SCENARIOS. In the first one, the
system generates a virtual train and projects it into the
augmented reality perceived by the test car as the train
approaches a mock-up of a rail crossing in Mcity. The point
is to see whether the test car can stop in time and then wait
for the train to pass. We also throw in other virtual vehicles,
such as cars that follow the test car. These strings of cars—
actual and ­virtual—can be formally arranged convoys (known
as ­platoons) or ad hoc arrangements: perhaps cars queuing
to get onto an entry ramp.
The second, more complicated testing scenario involves
the case we mentioned earlier—running a red light. In the
United States, cars running red lights cause more than a
quarter of all the fatalities that occur at an intersection,
according to the American Automobile Association. This
scenario serves two purposes: to see how the test car
reacts to traffic signals and also how it reacts to red-light-
running scofflaws.
Our test car should be able to tell whether the signal is
red or green and decide accordingly whether to stop or to
go. It should also be able to notice that the simulated red-
light runner is coming, predict its trajectory, and calculate
when and where the test car might be when it crosses that
trajectory. The test car ought to be able to do all these things
well enough to avoid a collision.
Because the computer running the simulation can fully
control the actions of the red-light runner, it can generate a
wide variety of testing parameters in successive iterations of
the experiment. This is precisely the sort of thing a computer
can do much more accurately than any human driver. And
of course, the entire experiment can be done in complete
safety because the lawbreaker is merely a virtual car.
There is a lot more of this kind of edge-case simulation
that can be done. For example, we can use the augmented-
reality environment to evaluate the ability of test cars to
handle complex driving situations, like turning left from
a stop sign onto a major highway. The vehicle needs to
seek gaps in traffic going in both directions, meanwhile
watching for pedestrians who may cross at the sign. The
car can decide to make a stop in the median first, or instead
simply drive straight into the desired lane. This involves
a ­d ecision-making process of several stages, all while
taking into account the actions of a number of other vehi-
cles (including predicting how they will react to the test
car’s actions).
Another example involves maneuvers at roundabouts—
entering, exiting, and negotiation for position with
other cars—without help from a traffic signal. Here the test
car needs to predict what other vehicles will do, decide on
an acceptable gap to use to merge, and watch for aggres-
sive vehicles. We can also construct augmented-reality
scenarios with bicyclists, pedestrians, and other road users,
such as farm machinery. The less predictable such alter-
native actors are, the more intelligence the self-­driving car
will need.
Ultimately, we would like to put together a large library of
test scenarios including edge cases, then use the augmented-
reality testing environment to run the tests repeatedly. We
are now building up such a library with data scoured from
reports of actual crashes, together with observations by
sensor-laden vehicles of how people drive when they don’t
know they’re part of an experiment. By putting together
disparate edge conditions, we expect to create artificial
edge cases that are particularly challenging for the soft-
ware running in self-driving cars.
Thus armed, we ought to be able to see just how safe a
given autonomous car is without having to drive it to the
sun and back. n
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/robocars1219
SPECTRUM.IEEE.ORG | DEC 2019 |  27
28  | DEC 2019 | SPECTRUM.IEEE.ORG
 M I CRO G R I D
I N A B OX
A plug-and-play microgrid will
make rooftop solar grid-friendly
A few years ago, I found
A myself in the laundry room
of a house in Austin, Texas,
looking at some disturbing
electrical signals on an oscilloscope. In
my capacity as chief technology officer
for the nonprofit clean-energy-research
firm Pecan Street, examining the effect
on the grid of homes with rooftop photo­
voltaics is my job. But what I was seeing
that day in the house’s connection to the
grid sparked an idea. “If I added energy
storage to this house, I could fix this,” I
remember saying.
If only it were that simple. Rooftop solar
seems like it should be straightforward.
When the sun shines, you should be able
to reduce or eliminate your home’s input
from the grid and maybe even sell some of
your power back to your utility company.
PHOTO-ILLUSTRATION BY Lincoln Agnew
BY S C O T T H I N S O N
With a home b ­ attery-storage unit, you
should be able to do that even after dark.
But the picture is more complicated
from the utilities’ perspective. Quite apart
from their worries that solar will reduce
their ability to pay for grid upkeep, resi-
dential rooftop solar has the potential to
amplify a problem that already exists in
most U.S. homes.
At the heart of the problem are switch-
mode power supplies—the kind that drive
most small appliances, entertainment sys-
tems, lighting, computers, mobiles, and
other devices. If these power supplies are
not painstakingly designed—and some of
them are not—they can have a corrupting
effect on the quality of the power on the
grid. On their own, the effect of a house,
or even a whole neighborhood, is minimal.
But add in grid-connected rooftop solar
and the quality problem has the potential
to strain the resources of the local distri-
bution network, as was shown in a 2015
U.S. Department of Energy (DOE) study
on integration of renewables in Texas.
In investigating the specific problems
in that customer’s house, we decided
that the best way to solve them was
with something more far-reaching
than just energy storage: a residential
microgrid in a box. It’s a system that
turns the user’s photovoltaics into a
reliable source of high-quality power,
not just for the home’s own use but for
the neighborhood and even the wider
grid. We weren’t the first to try this solu-
tion, and we certainly hope we won’t
be the last. Because we’ve come to
believe that without something like it,
or a change to consumer power-supply
requirements, residential rooftop solar Availability isn’t the only concern earlier, add another wrinkle—harmonic
will simply never reach anything like its regarding today’s residential solar power distortion. Here, voltage and current
full potential. systems. Another is their impact on aren’t just out of phase with each other;
power quality. Power quality is a catch- they don’t even share the same frequency.
all term for the effect on the grid of the All of these effects can be summed up
To understand why residen- devices plugged into it. It’s measured in in one term—power factor. It’s basically a
tial rooftop solar demands terms of frequency stability, voltage sta- measure of how much current on a line is
T a microgrid, you need a bit bility, and harmonic distortion of voltage doing work versus the capacity of the line.
of background on the elec- and current. The best way to explain it is with a cou-
tricity distribution grid, how the flow of To understand power quality, start with ple of examples. When the current in the
power on it is managed, the importance the fact that, ideally, voltage and cur- load is perfectly in phase with the voltage
of the quality of that power, and how rent oscillate in phase with each other and there is no distortion in the current,
today’s residential solar works. (at 60 hertz in North America and the the power factor (PF) is 1.0. As harmonic
Let’s start with the grid. From the begin- Caribbean, and at 50 Hz in much of the distortion increases or the phase angle of
ning of the electricity industry more than rest of the world). Simple resistive loads, the current starts to lead or lag the voltage,
a century ago, all power consumed on the like old-school incandescent lightbulbs, the power factor starts to drop. A PF of
grid was generated by a limited number of don’t affect that perfect synchrony. But 0.7, for example, might have a current that
plants. The output of these facilities was, more complex ones with capacitance or leads the voltage by about 45 degrees but
and still is, controlled by a hodgepodge of inductance, such as an air conditioner’s without any distortion. A PF of 0.5 could
regulatory and market regimes. The sys- compressor, can cause current to lead or have a leading current as well as signifi-
tem planning, modeling, and control of lag voltage. The farther the voltage and cant total harmonic distortion.
the grid was relatively uncomplicated up current slip out of phase, the less usable As the power factor drops and reactive
until about two decades ago because the power is delivered. power increases, the utility has to make
flow of power was always in one direction— Think of the situation this way: Power up for the change by generating compen-
from a utility-controlled generator to the is the product of voltage and current. satory reactive power to shift current and
residential load. There was little that hap- When they are perfectly in phase, the voltage back into phase with each other.
pened in your house that could alter that. products of both humps of the AC wave Without such support, voltage on the grid
With distributed generation, power are positive, meaning that power flows can sag and lead to blackouts. Histori-
sometimes flows in the opposite direc- in only one direction, from the gener- cally, this compensatory reactive power
is almost always capacitive, because some
of the biggest loads in homes and particu-
larly in industrial facilities are inductive.
For example, the compressors in refrig-
erators and air conditioners are driven
by motors that have coils.
Traditionally, utilities have been able
to assume that a residential structure is
POWER FACTOR: The relationship between the voltage and current waveforms influences a reasonably high-quality load. Histori-
power factor, a measure of how much current on a line is doing work versus the capacity of cal measurements on individual appli-
the line. In the ideal situation, the two waveforms are perfectly in phase with each other [left].
Inductive or capacitive loads can cause the current to lead or lag the voltage, reducing the ances and homes have borne this out,
power factor [center]. Harmonic distortion [right], which occurs when current flows at different with power factors of 0.9 or better and
frequencies than the voltage, reduces the power factor even further. relatively low harmonic distortion.
However, as more and more lighting
tion. But the utility on the receiving end ator to the load. All the power is the and electronics have moved to switch-
doesn’t have high confidence in how useful type, called active power. As mode power supplies, the power factor
much solar generation will be delivered current slips farther and farther out of of the typical home has dropped and the
at any particular time. Variable cloud phase with voltage, the active power reactive power support requirements
cover and changing seasons mean that decreases until it reaches zero, where have increased. Switch-mode supplies
from the utility’s perspective, residential the current waveform is 90 degrees out work by first rectifying the voltage from
solar compares poorly with organized of phase with voltage. Power is then the AC mains and then chopping it up into
central generation, where production 100 percent reactive, flowing in equal a high-frequency signal that allows for the
is controlled, reliable, and predict- amounts in both directions and doing use of relatively small transformers and
able. Nevertheless, grid operators have no useful work. other components. This signal’s voltage is
learned to handle such weather-related Certain loads, such as those using the then stepped down, rectified, and filtered
vagaries. switch-mode power supplies I mentioned to produce the desired direct-current out-

30  | DEC 2019 | SPECTRUM.IEEE.ORG

The power grid must
handle distortions
and spikes of
power caused by
appliances and
electronics.
ENERGY
SWITCH IN
ACTION:
During the day, as different
appliances are turned on and
off, the house draws varying
amounts of power [green].
These appliances reduce the
home’s power factor [orange]
by shifting current and voltage
out of phase or by distorting
the current waveform. The
power factor decreases even
more as the home’s solar array
Kilowatts
begins to produce power
[center]. The grid must be able
to compensate for both phase
and distortion. The Energy
Switch—the home’s energy-
storage and microgrid system—
limits the biggest power draws
[purple]. The system also takes
some of the burden from the
grid by increasing the power
factor. As a result, the home
acts as either a simple load
[yellow] or a power generator.
put at lower voltage. Unfortunately, those
high frequencies in the power supply can
show up as spikes of current in the mains,
contributing to harmonic distortion.
Traditional residential-grid-tied solar
installations in many regions are pretty
simple systems that don’t provide any
reactive power compensation. This
means that the solar array will allow any
phase displacement or distortion from
the home’s load right through to the util-
ity without any reduction—essentially
ILLUSTRATION BY MCKIBILLO
Electronics and
lighting with
switch-mode
power supplies
cause distortions
in current.
The home microgrid and energy-
storage system corrects for Appliances can reduce
distortions and power spikes the power factor by
while managing input from the pushing current and
rooftop solar array. voltage out of phase
with each other.
Solar being produced
1
6
0.5
4
Power factor
2 0
0
–0.5
–2
–1
–4
24 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
SOURCE: PECAN STREET Time
amplifying the problem. IEEE Standard electricity, the system provides current at
1547-2018 and regional mandates such as the correct 60-Hz fundamental frequency,
California’s Rule 21 Interconnection Tariff but it doesn’t do anything to cancel out
require solar installations that can help the harmonic distortion from the power
support the grid by producing reactive supplies running some of the appliances.
power; however, they do not solve any So the house at times produces more than
of the harmonic distortion issues. 100 percent total harmonic distortion
Consider a modern house in North at the grid connection. In that situation,
America with rooftop solar panels, a large the home is supplying more current at
central air conditioner, and several other harmonic frequencies than it does at the
common appliances and electronics. 60-Hz frequency that the distribution grid
When the rooftop panels are producing was designed to accommodate.
SPECTRUM.IEEE.ORG | DEC 2019 |  31

When only a few houses on the grid
present this type of load, it’s not that big
of an issue. But more solar arrays mean
more challenges for the utility. This was
shown in the previously mentioned DOE
study, which examined the power on feed-
ers to two neighborhoods in Texas. One
had rooftop solar and the other didn’t;
otherwise they were pretty similar. The
feeder to the solar neighborhood saw har-
monic distortions that regularly drove the
power factor below 0.65, while the non-
solar neighborhood’s power factor never
fell below 0.94. For solar to safely spread,
we need a better way.
At Pecan Street, we special-
A ize in data collection and
field trials for renewables to
reduce carbon, so we felt we
were in a good position to come up with
a solution. Working with the technology
development firm Concurrent Design,
also in Austin, in cooperation with the
DOE’s SunShot program, we designed
and built a residential energy-storage sys-
tem that turns a home into its own
microgrid. It’s a solution that addresses
all of solar’s integration issues at once.
The goals were to improve power quality
so that utilities can be more comfortable
accepting energy from residential solar
installations; to manage residential bat-
teries so that consumers can use energy
or deliver it to the grid even when the sun
32  | DEC 2019 | SPECTRUM.IEEE.ORG
SOLAR TOWN: Utilities want to know what
effect rooftop solar will have on the grid
so that they can compensate for it. Some
neighborhoods in Austin, Texas, already have
high solar penetration.
isn’t shining; and to allow homes to break
off from the grid and power themselves
during outages without so much as a
flicker during the transition. This combi-
nation, we believe, will enable higher pen-
etration of solar and other distributed
resources on existing grids designed for
traditional, centralized generation; it will
also make these distributed sources an
asset for utilities instead of the potential
operational liabilities they are now.
We call this system the Energy Switch.
We built a total of four of these systems,
two for lab testing and two for field tri-
als. The two field units were installed
in the homes of some very gracious vol-
unteers, who gave up a sizable chunk
of their garages for the project. Even
with the advances in energy storage, the
Energy Switch isn’t small—it’s roughly
the size of a tall, skinny home refrig-
erator at 61 centimeters wide, 61 cm
deep, and 180 cm tall. And it’s rather
heavy, at almost 900 kilograms (close to
2,000 pounds). Still, it’s a giant improve-
ment over our first energy-storage sys-
tem, developed six years ago, which
weighed in at 1,360 kg and was three
times the size.
The homes for field trials were about
220 square meters (close to 2,400 square
feet) and were built in an area of Austin
where the developer had specified aggres-
sive energy-efficiency measures. Both
homes had solar arrays connected directly
to the grid and rated at about 5,500 watts,
as well as heating, ventilation, and air-
conditioning (HVAC) units. One of the
homes had a plug-in electric vehicle, and
the other had an electric oven. The Energy
Switch was used to power the homes in
September and October of 2016. With
daytime temperatures often over 38 °C
(100 °F), air-conditioning usage is still
quite high in Austin during those months.
Initially we gave the Energy Switch
five operating modes: 1) maximize
­self-consumption of solar production;
2) maximize solar energy delivery to the
utility grid; 3) operate off-grid without
secondary generator support; 4) oper-
ate off-grid with secondary generator
support; and 5) system bypass: connect
directly to the grid for system mainte-
nance or battery-lifetime preservation.
Clearly, we were thinking as engineers,
not consumers. We thought these options
would give homeowners the flexibility
they wanted. But in hindsight, we proba-
bly should have simplified it to two visible
operating modes, or maybe even just one.
And we should have left the other choices,
such as self-consumption or grid produc-
tion, for the system to decide based on
grid needs and conditions. When we
explained these operating modes, one
of the participants’ children caught on
faster than his parents:
Him: “You mean…if the power
goes out I can still play video
games and have solar?”
Us: “Yes, and you’ll still have
air-conditioning…”
Him: “Sweet!!”
Getting to “Sweet!” was an enormous
amount of work that had to be done very
quickly. The total DOE grant period was
12 months, and we needed to reserve
three months at the end for demonstra-
tion, testing, and reporting. That left
just nine months to design and build
these systems.
PECAN STREET (2)
We did have access to some legacy hard-
ware that we’d previously developed.
However, to perform all of the power­
 factor correction, load control, genera-
tion control, and grid-support services we
envisioned would require a substantial
revamping of the entire system design.
We wanted the Energy Switch to
go beyond what other systems—even
new commercial systems like Tesla’s
­Powerwall 2 AC—could do. We wanted
the Energy Switch to enable a home to
push solar energy onto the grid with or
without a full battery and have full-time
power-­factor correction. We also wanted
the system to be fully integrated with the
home’s other energy systems so that it
would be able, for example, to accept
energy from a backup generator in addi-
tion to solar, to keep the house going in
an extended blackout.
To meet these challenges, we designed a
system that could monitor and control 24
different load-bearing circuits in the home.
The system makes about 300 decisions per
minute, configuring its circuitry based on
tens of thousands of measurements of the
batteries, the grid, the quality of its elec-
trical output, solar production, and the
loads in the home. These decisions make
sure that all systems are operating at their
highest possible efficiency and the home-
owner and the utility are getting the most
they can out of the rooftop solar and the
power-quality support system.
Of course, any energy-storage system
must include comprehensive safety mea-
sures. We designed the system so that
any of a wide variety of battery or elec-
tronics mishaps would result in a system
shutdown long before the fault reached a
critical level. That was tested on the first
day of field trials, when a bad capacitor—
easily fixed—in one home’s HVAC trig-
gered a surge of current that shut the
entire system down with a loud and dis-
maying relay clunk.
In its two months of service,
the Energy Switch proved
I its effectiveness in correct-
ing the power- qu alit y
issues. It kept harmonic distortion close
to a harmless and constant 20 percent,
preventing the wild swings in distortion
that a home can cause. And it reduced
reactive power so that the homes’ power
MICROGRID IN A BOX: The Energy Switch
combines 5,500 watts of energy storage with
circuits that limit the power draw from the grid
and also correct for distortion produced by
appliances and electronics in the home. The
system also manages power from rooftop solar
and backup generators when needed.
factor stayed quite close to 1.0 when pow-
ered by the utility.
When the t wo homes generated
enough power to put it back on the
grid, the Energy Switch ensured that
the power going to the grid had only a
minor distortion-­related reactive com-
ponent. So from the utility’s perspective,
the homes were now either generators or
resistive loads, making the job of voltage
control a lot easier for the utility.
Lastly, the system also provided pro-
grammable power regulation, limiting
the amount of power put back onto the
grid and the power drawn from the grid.
This is helpful for areas where growing
populations are straining utility distri-
bution capacity. In this case the Energy
Switch was set to limit the draw from the
grid to 3.5 kilowatts and generation to
0.5 kW. Without the Energy Switch the
homes would have occasionally drawn
and generated more than 4 to 5 kW.
The Energy Switch successfully dem-
onstrated many advanced functions—
functions beyond the scope of typical
energy-storage systems. These functions
and operating capabilities will need to
become more commonplace if residen-
tial solar is to become a significant part
of the grid, both because they improve
utility operations and because consum-
ers want them.
With the Energy Switch, consumers
get a reliable power source, savings on
their utility bill, and a degree of immu-
nity to blackouts. While all those fea-
tures have benefits you can count in
dollars and cents, the last also gives a
palpable sense of security. After the trial
was over, both families in our Austin
experiment said they missed the feel-
ing of security that the Energy Switch
gave them, even though the utility didn’t
have a single outage in the neighbor-
hood during the trial.
That said, we designed a system that
the volunteer participants wanted, not
necessarily what they could afford.
When we interviewed potential volun-
teers for field trials, we discussed bat-
tery sizing options. In the interviews we
conducted, people almost universally
rejected any storage solution that would
run only a small subset of their appli-
ances. Once the idea of battery backup
for the whole home became a possibil-
ity, participants quite definitely wanted
that functionality—though, predictably,
they were wary of the cost. At the time
the project ended we estimated that
whole-home battery systems like the
Energy Switch could be sold for about
US $10,000 to $20,000, depending on
features, battery size, and advancements
in electronics integration. Admittedly,
with costs like that, systems such as the
Energy Switch aren’t for everyone. But
we hope that at least it will point to a
more affordable solution.
From a utility’s perspective, combin-
ing photovoltaics, storage, and a home
microgrid turns residential solar into
something that takes less active manage-
ment. And that means utilities have fewer
reasons to want to restrict the spread of
rooftop solar. Fearing instability, utili-
ties have sometimes capped the portion
of residential solar systems they’ll allow
to be tied to a neighborhood feeder line
at 30 percent or even lower. Why can’t
it be 100 percent? We think residential
microgrids can help us get there. n
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/
microgrid1219
By Leaps and Bounds
An exclusive look at how
Boston Dynamics
is redefining robot agility
By ERICO GUIZZO • Photography by BOB O’CONNOR

34  | DEC 2019 | SPECTRUM.IEEE.ORG
WWITH THEIR JAW-DROPPING agility and animal-like reflexes,
Boston Dynamics’ bioinspired robots have always seemed
to have no equal. But that preeminence hasn’t stopped
the company from pushing its technology to new heights,
sometimes literally. Its latest crop of legged machines can
trudge up and down hills, clamber over obstacles, and even
leap into the air like a gymnast. There’s no denying their
appeal: Every time Boston Dynamics uploads a new video
to ­YouTube, it quickly racks up millions of views. These
are probably the first robots you could call Internet stars.
Boston Dynamics, once owned by Google’s parent company,
­Alphabet, and now by the Japanese conglomerate SoftBank,
has long been secretive about its designs. Few publications
have been granted access to its Waltham, Mass., headquarters,
in action—running, climbing, jumping—by using high-speed
cameras coupled with powerful strobes. The results you see
on these pages: freeze-frames of pure robotic agility.
We also used the photos to create interactive views,
which you can explore online on our Robots Guide (https://
robots.ieee.org). These interactives let you spin the robots
360 degrees, or make them walk and jump on your screen.
Boston Dynamics has amassed a minizoo of robotic
beasts over the years, with names like BigDog, SandFlea,
and ­WildCat. When we visited, we focused on the two most
advanced machines the company has ever built: Spot, a
­nimble ­quadruped, and Atlas, an adult-size humanoid.
Spot can navigate almost any kind of terrain while sensing
its environment. Boston Dynamics recently made it avail-
able for lease, with plans to manufacture something like a
thousand units per year. It envisions Spot, or even packs of
them, inspecting industrial sites, carrying out hazmat mis-

near Boston. But one morning this past August, IEEE ­Spectrum
got in. We were given permission to do a unique kind of photo
shoot that day. We set out to capture the company’s robots
In the Spotlight
BOSTON DYNAMICS designed Spot as a ­versatile
mobile machine suitable for a variety of applications.
The company has not announced how much Spot
will cost, saying only that it is being made available to select
­customers, which will be able to lease the robot. A payload bay
lets you add up to 14 kilograms of extra hardware to the robot’s
back. One of the accessories that Boston Dynamics plans to
offer is a 6-degrees-of-freedom arm, which will allow Spot to
grasp objects and open doors.

SPECTRUM.IEEE.ORG | DEC 2019 |  35

sions, and delivering packages. And its YouTube fame
has not gone unnoticed: Even entertainment is a pos-
sibility, with Cirque du Soleil auditioning Spot as a
potential new troupe member.
“It’s really a milestone for us going from robots that
work in the lab to these that are hardened for work
out in the field,” Boston Dynamics CEO Marc Raibert
says in an interview.
Our other photographic subject, Atlas, is Boston
Dynamics’ biggest celebrity. This 150-centimeter-tall
(4-foot-11-inch-tall) humanoid is capable of impres-
sive athletic feats. Its actuators are driven by a com-
pact yet powerful hydraulic system that the company
engineered from scratch. The unique system gives the
80-kilogram (176-pound) robot the explosive strength
needed to perform acrobatic leaps and flips that don’t
seem possible for such a large humanoid to do. Atlas
has inspired a string of parody videos on YouTube and
more than a few jokes about a robot takeover.
While Boston Dynamics excels at making robots, it
has yet to prove that it can sell them. Ever since its
founding in 1992 as a spin-off from MIT, the company
has been an R&D-centric operation, with most of its
early funding coming from U.S. military programs.
The emphasis on commercialization seems to have
intensified after the acquisition by SoftBank, in 2017.
­SoftBank’s founder and CEO, Masayoshi Son, is known
to love robots—and profits.
The launch of Spot is a significant step for Boston
­Dynamics as it seeks to “productize” its creations. Still,
Raibert says his long-term goals have remained the
same: He wants to build machines that interact with
the world dynamically, just as animals and humans do.
Has anything changed at all? Yes, one thing, he adds
with a grin. In his early career as a roboticist, he used
to write papers and count his citations. Now he counts
YouTube views. n
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/bostondynamics1219
36  | DEC 2019 | SPECTRUM.IEEE.ORG
Spot
HEIGHT WEIGHT SPEED
84 cm 25 kg 5.76 km/h
SENSING: Stereo cameras, ACTUATION: 12 DC motors
inertial measurement unit, POWER: Battery
position/force sensors (90 minutes per charge)
Super Senses
SPOT’S HARDWARE
is almost entirely custom-
designed. It includes powerful
processing boards for control as well
as sensor modules for perception. The
­sensors are located on the front, rear, and
sides of the robot’s body. Each module
consists of a pair of stereo cameras,
a wide-angle camera, and a texture
projector, which enhances 3D sensing
in low light. The sensors allow the robot
to use the navigation method known as
SLAM, or simultaneous localization and
mapping, to get around autonomously.
Stepping Up
IN ADDITION TO
its autonomous behaviors, Spot
can also be steered by a remote
operator with a game-style controller. But
even when in manual mode, the robot
still exhibits a high degree of autonomy.
If there’s an obstacle ahead, Spot will go
around it. If there are stairs, Spot will climb
them. The robot goes into these operating
modes and then performs the related
actions completely on its own, without
any input from the operator. To go down
a flight of stairs, Spot walks backward, an
approach Boston Dynamics says provides
greater stability.
Funky Feet
SPOT’S LEGS are powered by 12 custom DC
motors, each geared down to provide high
torque. The robot can walk forward, sideways, and
backward, and trot at a top speed of 1.6 meters per second.
It can also turn in place. Other gaits include crawling and
pacing. In one wildly popular YouTube video, Spot shows off
its fancy footwork by dancing to the pop hit “Uptown Funk.”

SPECTRUM.IEEE.ORG | DEC 2019 |  37
Atlas
HEIGHT WEIGHT

150 cm 80 kg
SPEED

5.4 km/h
SENSING:
Lidar and stereo vision
ACTUATION:
28 hydraulic actuators
POWER: Battery

Robot Blood
ATLAS IS POWERED
by a hydraulic system
consisting of 28 actuators.
These actuators are basically
cylinders filled with pressurized fluid
that can drive a piston with great
force. Their high performance is
due in part to custom servo valves
that are significantly smaller and
lighter than the aerospace models
the company had been using in
earlier designs. Though not visible
from the outside, the innards of an
Atlas are filled with these hydraulic
actuators as well as the lines of fluid
that connect them. When one of
those lines ruptures, Atlas bleeds
the hydraulic fluid, which happens
to be red.

Biologically Inspired
ATLAS’S CONTROL software
doesn’t explicitly tell the robot
how to move its joints, but rather
it employs mathematical models of the
underlying physics of the robot’s body and
how it interacts with the environment. Atlas
relies on its whole body to balance and move.
When jumping over an obstacle or doing
acrobatic stunts, the robot uses not only its
legs but also its upper body, swinging its
arms to propel itself just as an athlete would.

38  | DEC 2019 | SPECTRUM.IEEE.ORG
Next Generation
THE CURRENT VERSION of Atlas is a
thorough upgrade of the original model,
which was built for the DARPA Robotics
Challenge in 2015. The newest robot is lighter and
more agile. Boston ­Dynamics used ­industrial-grade
3D printers to make key structural parts, giving the robot
greater strength-to-weight ratio than earlier designs.
The next-gen Atlas can also do something that its
predecessor, famously, could not: It can get up after a fall.

Walk This Way

TO CONTROL ATLAS, an
operator provides general
steering via a manual
controller while the robot uses its stereo
cameras and lidar to adjust to changes
in the environment. Atlas can also
perform certain tasks autonomously.
For example, if you add special bar-
code‑type tags to cardboard boxes,
Atlas can pick them up and stack
them or place them on shelves.

SPECTRUM.IEEE.ORG | DEC 2019 |  39
THAR SHE BLOWS:
The SnotBot drone
passes over a blue
whale at the moment
of exhalation [top left].
A humpback whale
rolls as a drone
approaches [top
right]. SnotBot
passes over a
surfacing humpback
whale off the coast
of Gabon, in Africa
[bottom right]. The
drone approaches a
blue whale mother
and calf in the
Gulf of California
[bottom left].

40  | DEC 2019 | SPECTRUM.IEEE.ORG
SNOTBOT:
A Whale of a Deep-Learning Project

Drones are collecting snot and images

to monitor whale health
by b ry n k e l l e r & t e d w i l l k e

SPECTRUM.IEEE.ORG | DEC 2019 |  41
It’s a beautiful morning on the waters of Alaska’s Peril Strait—clear, calm, silent, and just a little
cool. A small but seaworthy research vessel glides through gentle swells. Suddenly, in the distance, a
humpback whale the size of a school bus explodes out of the water. Enormous bursts of air and water
jet out of its blowholes like a fire hose, the noise echoing between the banks. • “Blow at eleven o’clock!”
cries the lookout, and the small boat swarms with activity. A crew member wearing a helmet and
cut‑proof gloves raises a large quadcopter drone over his head, as if offering it to the sun, which glints
off the half dozen plastic petri dishes velcroed to the drone. • Further back in the boat, the drone pilot
calls, “Starting engines in 3, 2, 1! Takeoff in 3, 2, 1!” The drone’s engines buzz as it zooms 20 meters into PREVIOUS PAGES AND THESE PAGES: CHRISTIAN MILLER/OCEAN ALLIANCE (8)

the air and then darts off toward where the whale just dipped below the water’s surface. With luck, the
whale will spout again nearby, and the drone will be there when it does.

The drone is a modified DJI Inspire 2.
About the size of a toaster oven, it’s gen‑
erally sold to photographers, cinema‑
tographers, and well-heeled hobbyists,
but this particular drone is on a serious
mission: to monitor the health of whales,
the ocean, and by extension, the planet.
The petri dishes it carries collect the
exhaled breath condensate of a whale—
a.k.a. snot—which holds valuable infor‑
mation about the creature’s health, diet,
and other qualities. Hence the drone’s
name: the Parley SnotBot.
The flyer comes standard with a
­forward-facing camera for navigation,
collision-avoidance detectors, ultra‑
sonic and barometric sensors to track
altitude, and a GPS locator. With the
addition of a high-definition video cam‑
era on a stabilized gimbal that can be
directed independently, it can stream
1080p video live while simultaneously
storing the video on a microSD card
as well as high-resolution images on a
1-terabyte solid-state drive. Given that
both cameras run during the entire
26 minutes of a typical flight, that’s a
lot of data. More on what we are doing
with that data later, but first, a bit of
­SnotBot history.

42  | DEC 2019 | SPECTRUM.IEEE.ORG
 SNOT ON BOARD: In the photos at left, coauthor Bryn Keller, holding the
drone, and Iain Kerr, the chief executive officer of Ocean Alliance, retrieve
petri dishes that were sprayed with whale exhalation. On this page, Kerr,
sitting under the awning, and other researchers send SnotBot on a hunt
for whales in the Gulf of California.

Iain Kerr was one of the early pioneers
in using drones as a platform to collect
and analyze whale exhalation. He’s the
CEO of Ocean Alliance, in Gloucester,
Mass., a group dedicated to protecting
whales and the world’s oceans. Whale
biologists know that whale snot con-
tains an enormous amount of biological
information, including DNA, hormones,
and microorganisms. Scientists can use
that information to determine a whale’s
health, sex, and pregnancy status, and
details about its genetics and microbiome.
The traditional and most often used tech-
nique for collecting that kind of informa-
tion is to zoom past a surfacing whale in a
boat and shoot it with a specially designed
crossbow to capture a small core sample
of skin and blubber. The process is stress-
ful for both researchers and whales.
Researchers had demonstrated that
whale snot can be a viable replacement for
blubber samples, but collection involved
reaching out over whales using long, awk-
ward poles—difficult, to say the least. The
development of small but powerful com-
mercial drones inspired Kerr to launch an
exploratory research project in 2015 to go
after whale snot with drones. He received
the first U.S. National ­Oceanic and Atmo-
spheric Administration (NOAA) research
permit for collecting whale snot in U.S.
waters. Since then, there have been doz-
ens of SnotBot missions around the world,
in the waters off Alaska, Gabon, Mexico,
and other places where whales like to
congregate, and the idea has spread to
other teams around the globe.
The SnotBot design continues to evolve.
The earliest versions tried to capture snot
by trailing gauzy cloth below the drone.
The hanging cloth turned out to be dif-
ficult to work with, however, and the
material itself interfered with some of
the lab tests, so the researchers scrapped
that method. The developers didn’t con-
sider using petri dishes at first, because
they assumed that if the drone flew
directly into a whale’s spout, the rotor
wash would interfere with collection.
Eventually, though, they tried the petri
dishes and were happy to discover that
the rotors’ downdraft improved rather
than hindered collection.
For each mission, the collection goals
have been slightly different, and the team
tweaks the design of the craft accordingly.
On one mission, the focus might be to
survey an area, getting samples from as
many whales as possible. The next mis-
sion might be a “focal follow,” in which
the team tracks one whale over a period
of hours or days, taking multiple samples
so that they can understand things like
how a whale’s hormone levels change
throughout the day, either from natural
processes or as a response to environ-
mental factors.
C ollecting and analyzing snot
is certainly an important way to assess
whale health, but the SnotBot team sus-

SPECTRUM.IEEE.ORG | DEC 2019 |  43
pected that the drone could do more. In
early 2017, staffers from Parley for the
Oceans, a nonprofit environmental group
that was working with Ocean ­Alliance on
the SnotBot project, contacted one of us
(Willke) to find out just how much more.
Willke is a machine-learning and
­artificial-intelligence researcher who
leads Intel’s Brain-Inspired Computing
Lab, in ­Hillsboro, Ore. He immediately
saw ways of expanding the information
gathered by SnotBot. Willke enlisted two
researchers in his lab—coauthor Keller
and Javier Turek—and the three of us got
to work on enhancing SnotBot’s mission.
The quadcopters used in the ­SnotBot
project carry high-quality cameras with
advanced auto-stabilization features. The
drone pilot relies on the high-definition
video being streamed back to the boat
to fly the aircraft and collect the snot.
We knew that these same video streams
could simultaneously feed into a com-
puter on the boat and be processed in
real time. Could that information help
assess whale health?
Working with Ocean Alliance scientists,
we first came up with a tool that ana-
lyzes a photo of a whale’s tail flukes and,
using a database of whale photographs
collected by the Alaska Whale Founda-
tion, identifies individual whales by the
shape of the fluke and its black and white
patterns. Identifying each whale allows
researchers to correlate snot samples
over time.
Such identification can also help whale
biologists cope with tricky regulatory
issues. For example, there are at least
two breeding populations of humpback
whales that migrate to Alaska. Most come
from Hawaii, but a smaller group comes
from Mexico. The Mexican population is
under greater stress at the moment, and
44  | DEC 2019 | SPECTRUM.IEEE.ORG
so NOAA requests that researchers focus
on the healthier, Hawaiian whales and
leave the Mexican whales alone as much
as possible. However, both popu­lations
are exactly the same species and thus
indistinguishable from each other as a
group. The ability to recognize individual
whales allows researchers to determine
whether a whale had been previously
spotted in Mexico or Hawaii, so that they
can act appropriately to comply with
the regulation.
We also developed software that ana-
lyzes the shape of a whale from an over-
head shot, taken about 25 meters directly
above the whale. Since a skinny whale is
often a sick one or one that hasn’t been
getting enough to eat, even that simple
metric can be a powerful indicator of
well-being.
The biggest challenge in develop-
ing these tools was what’s called data
­s tarvation—there just wasn’t enough
data. A standard deep-learning algorithm
would look at a huge set of images and
then figure out and extract the key distin-
guishing features of a whale. In the case
of the fluke-ID tool, there were only a
few pictures of each whale in the catalog,
and these were often too low quality to
be useful. For overhead health monitor-
ing, there were likewise too few photos
or videos of whales shot with the right
camera, from the right angle, under the
right conditions.
To address these problems, our team
turned to classic computer-vision tech-
niques to extract what we considered
the most useful data. For example, we
used edge-detection algorithms to find
and measure the trailing edge of a fluke,
then obtained the grayscale values of
all the pixels in a line extending from
the center notch of the fluke to the outer
tips. We trained a small but effective neu-
ral network on this data alone. If more
data had been available, a deep-learning
approach would have worked better than
our approach did, but we had to work
with the limited data we had.
N e w d i s c o v e r i e s in whale biol-
ogy have already come from our tools.
Besides the ability to distinguish between
the Mexican and Hawaiian whale pop-
ulations, researchers have discovered
they can identify whales from their calls,
even when the calls were recorded many
years previously.
That latter discovery came during the
summer of 2017, when we joined Fred
Sharpe, an Alaska Whale Foundation
researcher and founding board member,
to study teams of whales that worked
together to feed. While observing a small
group of humpback whales, the boat’s
underwater microphone picked up a
whale feeding call. Sharpe thought it
sounded familiar, and so he consulted
his database of whale vocalizations. He
found a similar call from a whale called
Trumpeter that he had recorded some
20 years ago. But was it really the same
whale? There was no way to know for
sure from the whale call.
Then a whale surfaced briefly and
dove again, letting us capture an image
of its flukes. Our software found a match:
The flukes indeed belonged to T ­ rumpeter.
That told the researchers that adult
whale feeding calls likely remain stable
CHRISTIAN MILLER/OCEAN ALLIANCE (3)
for decades, maybe even for life. This
insight gave researchers another tool
for identifying whales in the wild and
improving our understanding of vocal
signatures in humpback whales.
Meanwhile, whale-ID tools are get-
ting better all the time. The original

­ notBot algorithm that we developed
S humpback and southern right whales,
for whale identification has been essen-
tially supplanted by more capable ser-
vices. One new algorithm relies on the
curvature of the trailing edge of the fluke
for identification.
SnotBot’s real contribution, it turns
out, is in health monitoring. Our shape-
analysis tool has been evolving and,
in combination with the spray samples,
is giving researchers a comprehen-
sive picture of an individual whale’s
health. We call this tool Morphometer.
We recently teamed up with Kelly Cates,
a Ph.D. candidate in marine biology at
ALL WET: The latest
model of SnotBot
[opposite page] flies
into action, with
custom mounting
points for petri dishes
KATE WESTAWAY/GETTY IMAGES
and its new paint
scheme, designed to
camouflage it against
a cloud-studded
sky. A humpback
whale dives below
the surface of the
ocean [right].
the University of Alaska Fairbanks, and
Fredrik Christiansen, an assistant profes-
sor and whale expert at the Aarhus Insti-
tute of Advanced Studies, in Denmark,
to make the technology more powerful
and also easier to use.
Here’s how it works. Researchers who
make measurements and assessments
of baleen whales—the type of whales
that filter-feed—have typically used a
technique developed by Christiansen
in 2016. (So far the effort has involved
but the process could work for any kind
of baleen whale.) The researchers start
with photographic prints or images
on a computer and hand-measure the
body widths of whales in the images
at intervals of 5 percent of the overall
length from the snout to the notch of
the tail flukes. They then feed this set of
measurements to software that calcu-
lates an estimate of the whale’s volume.
From the relationship between the body
length and volume, they can determine
if an individual whale is relatively fatter
or thinner compared with population
norms, taking into account the significant
but normal changes in girth that occur
as whales accumulate energy reserves
during the feeding season and then use
those energy stores for migration and
during the breeding season.
Morphometer also uses photos, but
it measures the whale’s width continu-
ously at the highest resolution possible
given the quality of the photo, yielding
hundreds of width measurements for
each animal, instead of only the small
number of measurements that are fea-
sible for human researchers. The result
is thus much more accurate. It also pro-
cesses the data much faster than a human
could, allowing biologists to focus on
biology rather than doing tedious mea-
surements by hand.
To improve Morphometer, we trained a
deep-learning system on images of hump-
back and southern right whales in all
sorts of different weather, water, and
lighting conditions to allow it to under-
stand exactly which pixels in an image
belong to a whale. Once a whale has
been singled out, the system identifies
CO NTI N U E D O N PAG E 53
How to
Shut Down
Robocallers
The STIR/
SHAKEN
protocol
will stop
scammers
from
exploiting
a caller ID By JIM
M c E AC H E R N
loophole & ERIC
BU RG E R

H
ave you ever received a phone call from your own number? If so, you’ve expe-
rienced one of the favorite techniques of phone scammers. • Scammers can
“spoof ” numbers, making it seem as though the phone call in question is com-
ing from a local number—which can include your own—thereby obscuring the
call’s true origin. If you answer the call, you’ll most likely be treated to the sound
of a robotic voice trying to trick you into parting with some money. • One of us
(McEachern) is a principal technologist for the standards organization Alliance
for Telecommunications Industry Solutions (ATIS), and the other (Burger) was
until recently the chief technology officer for the U.S. Federal Communications
Commission. But you don’t need us to tell you that robocalls are a pandemic.
According to a report by the caller ID company Hiya, there were 85 billion robo-
calls globally in 2018. →

46  | DEC 2019 | SPECTRUM.IEEE.ORG PHOTOGRAPH BY Dan Saelinger

GUTTER CREDIT GOES HERE

PHOTOGRAPH BY Firstname Lastname SPECTRUM.IEEE.ORG | DEC 2019 |  47

Call for Directions
CALLING A FRIEND isn’t
as simple as transmitting
data from point A to point
B. Along the way, the
call is routed through
telephone infrastructure
that may be operated by
two, three, or more phone
companies, or carriers.
These routes are part of
the reason it’s so time
consuming to identify the
points where scammers CALLER’S
place their calls. PHONE
RoboKiller, one company that has cre-
ated an anti-spam-call app, estimates
that Americans received 5.3 billion robo-
calls in April 2019 alone, or nearly 4,000
every second. And not only are scam calls
annoying, they’re costly. In 2018, phone
scams tricked Americans out of an esti-
mated US $429 million. Sadly, these num-
bers are on an upward trend.
Spoofing phone numbers is just one
way phone scammers trick their victims.
Scammers are also very good at reading
people, gaining their confidence, and
playing to their fears. But spoofing num-
bers is an often-effective opening gam-
bit. The first thing a spoofer has to do is
get someone to pick up the phone in the
first place, and people are more likely
to answer a call if they think it’s from a
local number. So, preventing the abuse of
call spoofing, along with making it much
harder for anyone to place huge numbers
of robocalls, are two of the most impor-
tant challenges to reining in robocallers
and scammers.
The telecommunications industry has
been developing a network-based system
that would meet both of these challenges.
It goes by an unwieldy name: “Secure Tele-
phone Identity Revisited/Signature-based
Handling of Asserted information using
toKENs.” Let’s just call it STIR/SHAKEN,
which is a lot easier to remember. STIR/
SHAKEN is a technique for providing more
reliable call-display information by clos-
ing a loophole that scammers exploit in
telephony infrastructure.
48  | DEC 2019 | SPECTRUM.IEEE.ORG
CENTURY AT&T
LINK LONG DISTANCE
AT&T
NODE
VERIZON INTELIQUENT
LONG DISTANCE
Today, when you make a call, your
phone company, or carrier, knows
whether or not you’re spoofing your num-
ber to make it appear that the call is com-
ing from a different number. But what the
company doesn’t know is if you’re allowed
to spoof that number, nor does it have a
way to securely send that information to
the carrier delivering the call to the person
you’re calling (there are legitimate reasons
why callers might spoof their numbers;
more on that in a moment).
The upshot is that when you see the
number of an incoming call, you have
no way of knowing if the number dis-
played on your caller ID is legitimate or
spoofed. STIR/SHAKEN will give phone
companies a secure method of commu-
nicating a caller’s number to a recipient
when a call is placed. This capability is
vital to establishing the caller’s reputation
so that scammers and other bad actors
can be reliably identified and blocked
before you waste any time on the bogus
call. And should an illegitimate robo-
call still get through, STIR/SHAKEN sim-
plifies the process of tracing a call back
to its source. Hopefully, simpler trac-
ing will make it feasible for law enforce-
ment agencies to prosecute scammers
for illegal robocalling. The technology
will also securely provide information to
call-blocking apps, allowing the apps to
more accurately identify spam calls and
inform you with a notice such as “Spam
likely” or “Unverified number” before
you answer a call.
VERIZON
RECEIVER’S
PHONE
If it all works out, robocalls could
become as manageable as email spam.
You’ll be less likely to be tricked into
answering a scam call, and you’ll receive
far fewer in the first place.
Spoofing numbers isn’t new—it’s
been possible for half a century. Tele-
phone switching equipment known as
private branch exchanges (PBXs), which
many businesses use, preassign the num-
ber that will be displayed on the recipi-
ent’s phone when it receives a call from
the business. There are legitimate rea-
sons for businesses to spoof numbers.
For example, they may want to display
a toll-free number for calls from the mar-
keting, sales, or service departments.
Women’s shelters are another example
of the need to disguise numbers, as they
often replace the shelter’s actual number
with a national number to avoid tipping
off a domestic abuser.
The problem is not spoofing itself. The
problem is that in the last decade or so,
three things have changed to create the
mess we see today.
First, phone calls are a lot cheaper. In
many countries nowadays, unlimited
nationwide calling is standard in basic
phone plans. Second, the Internet has
reduced the costs of running a scam to
almost nothing. The PBX of choice for
fraudsters is an Internet-enabled I­ P-PBX
to further lower the price per call. With
the Internet, scammers don’t even
have to be in the same country to place
ILLUSTRATION BY Mark Montgomery
robocalls, and they can use live agents
in countries with low-cost labor. Third,
anyone can place hundreds of calls per
minute with the small investment of an
inexpensive PC, a hundred-dollar Voice
over Internet Protocol (VoIP) expansion
card, free open-source software, and a
few days of assembly.
Put it all together and you have the rec-
ipe for a potentially lucrative business
with very low risk. These scammers are
fundamentally playing a numbers game:
While most people won’t answer their
calls, a small percentage will, and some of
those people can be conned into sending
money or revealing their bank account
information. Robocall scams are so cheap
that even one success among hundreds
or thousands of calls can still make scam-
mers money. The Internet can cheaply
connect a U.S.-based IP-PBX making hun-
dreds of calls per minute with call agents
in another country to talk to any victims
who fall for the spoofed call. Meanwhile,
the carrier has no way of know-
ing that this is an illegal robocall
operation until unsuspecting vic-
tims complain. Only after receiv-
ing complaints is the carrier aware
that it should trace back the calls
and identify the illegal caller.
Before STIR/SHAKEN, individual
phone companies did not have all
the information needed to iden-
tify and stop a scam, because it
often takes two or three compa-
nies to complete a call. The last
company in the chain, which com-
pletes the scammer’s connection
to your phone, doesn’t know if the
number has been illegally spoofed,
so it can’t advise you to use cau-
tion or ignore the call. The scam
emerges only when you answer
the call and discover that it isn’t
really the tax collector on the other
end of the line.
D e s p i t e t h e Ja m e s B o n d
theme, the STIR and SHAKEN
technologies don’t by themselves
constitute a license to kill robo-
calls. Instead, the goal is simply
to communicate, securely and in
real time, information between
PHOTOGRAPH BY Dan Saelinger
the phone companies on each end of
a phone call.
The Internet Engineering Task Force
(IETF), which works on issues related to
secure telephone identity, began work
on STIR in 2013. The IETF designed the
STIR protocol to be very flexible. The
basic mechanism is a certificate issued
to authenticated callers. However, STIR
requires individuals to be proactive about
authenticating themselves and managing
their personal key, which confirms their
identity. STIR’s downside is that very few
people have the expertise to do either. The
good news is that STIR’s flexibility allows
phone companies to implement it in their
network with minimal hassle.
In 2015, ATIS also began studying
mechanisms to reduce unwanted robo-
calls. A joint task force between ATIS
and the SIP Forum, an industry associa-
tion, built upon the IETF’s work on STIR.
As it turned out, STIR’s extreme flex-
ibility was a problem. Indeed, the pro-
tocol’s flexibility made it easy for each
phone company to implement it in its
network. However, as a general rule, the
more flexible a protocol is, the more
likely it is that different implementations
won’t play well together. So when two
different service providers implement
the protocol on each of their networks,
a caller ID sent from one to the other
might not make it through intact. The
task force’s goal was to create a precisely
defined subset (known as a profile) of
STIR, called SHAKEN. Because the task
force specified the SHAKEN profile of the
STIR protocol, you might see it referred
to as “STIR/SHAKEN.”
SHAKEN starts with the information
that the originating phone company—
the carrier—knows about the call. For
example, mobile phones and residen-
tial landlines transmit their phone num-
bers whenever they originate a call. For
businesses, where legitimate spoofing is
commonplace, the carrier also assigns
to the call a unique key, called
an “orig-id,” or origination identi-
fier, in order to identify the busi-
ness placing the call. In all cases,
the carrier creates a digital sig-
nature using the available infor-
mation and transmits it with the
call. The caller ID information is
included within this digital signa-
ture. The phone company com-
pleting the call verifies the digital
signature to confirm the infor-
mation hasn’t been modified,
and then identifies the originat-
ing carrier. This last step allows
spoofed calls to be linked to their
source for call-blocking apps and
law enforcement.
SHAKEN’s contribution is to
take what the originating phone
company knows about the caller,
courtesy of the digital signature,
and classify that knowledge suc-
cinctly. So one of the biggest chal-
lenges the task force faced early
on during SHAKEN’s develop-
ment was deciding which infor-
mation was actually important.
Including too little information
in the classification would mean
that important details would be
How STIR/SHAKEN
Tracks Down
- - - - - -
- - - - - -

a Scammer
- - - - - -
- - - - - -
- - - - - -
- - - - - -
When a carrier rolls out STIR/SHAKEN, the - - - - - -
only change its customers will notice is a - - - - - -
message on their caller ID screens warning of - - - - - -
a potential scam call. But there’s a lot more - - - - - -
going on behind the scenes when a scammer
places a robocall. Here’s how STIR/SHAKEN
keeps everyone involved informed about
whether a call is worth answering. 1. A scammer starts up the robocalling equipment and begins placing calls.

2. The scammer’s LEVEL 3 3. The carrier

own carrier logs also assigns an ALERT: SCAMMER!
each robocall’s “attestation level”
entry point—the (A, B, or C) to the
device used and its call based on what
physical location— the carrier knows
into the telephone about the caller.
network.

Attestation A

Attestation B

Attestation C

4. The carrier encrypts this information and sends it through the 5. Using the assigned attestation level, and taking into account
network, alongside the call itself, to the call receiver’s carrier. previous complaints about calls from the same network entry point,
the carrier determines the caller’s reputation as a likely scammer.

FCC
- - - - - -
- - - - - -
- - - - - -
- - - - - -
- - - - - -
- - - - - -
- - - - - -
- - - - - -
FCC - - - - - -
- - - - - -

6a. The call recipient 6b. If the recipient answers a scam 7. The recipient’s carrier, and the authorities, can trace the call
avoids picking up robocall, they can report the robocaller back to its origin using the entry point logged by the first carrier,
a phone call from a to their carrier and the authorities. allowing for prosecution.
probable scammer.

-
50  | DEC 2019 | SPECTRUM.IEEE.ORG ILLUSTRATION BY Mark Montgomery
lost, such as the distinctions among indi-
vidual businesses in a single building. Too
much information would create clutter
and make it more difficult to zero in on
the data that’s important. For example,
you don’t need to know whether a caller
is using a landline or a mobile phone
to determine whether they’re illegally
spoofing a number.
The solution was a three-level system
to categorize the essential information
about the caller into levels of “attesta-
tion” for the call. These attestation lev-
els characterize a caller’s right to use a
particular number. Full attestation, also
known as “A-attestation,” has several
requirements but provides the highest
level of confidence by the originating
carrier. The call originates on the car-
rier’s own network, as opposed to orig-
inating from another carrier or a VoIP
provider. The carrier has also directly
authenticated the caller and verified
the caller’s right to use the number. This
way, SHAKEN still allows for legitimate
number spoofing, but only if the car-
rier knows the customer has the right
to spoof that number.
Partial attestation, or “B-attestation,”
indicates that the originating carrier can-
not verify enough information about the
caller for the carrier to vouch that the
caller is using its assigned number. The
call still originates on the carrier’s net-
work. The carrier still authenticates the
caller but does not verify the customer’s
right to use the number that’s being dis-
played. It’s possible that the customer is
using the number legitimately, but the
carrier hasn’t verified it. There are valid
reasons why a customer might be using
a number that hasn’t been verified. A
business might have swapped carriers
but kept its original toll-free number,
for example.
Gateway attestation, or “C-attestation,”
indicates the lowest level of confidence.
The call starts on some other carri-
er’s network that hasn’t implemented
SHAKEN. Because the carrier doesn’t
know the customer or whether it has
the right to use the number that is asso-
ciated with the call, the carrier merely
identifies the call’s entry point into its
network. Gateway attestation may not
carry a lot of confidence about the call- are placed, SHAKEN makes it possible
er’s identification, but it can still be use- to confidently label a call as spam before
ful in tracing calls to quickly identify the it is answered.
source of problems.
Identifying a call as spam can happen One criticism of SHAKEN is that it
only once a call is placed. This limita- cannot indicate whether a call is a scam
tion highlights a key difference between based on whether or not the number is
phone calls and email, and helps explain legitimate. A call with “full attestation”
why we’ve had spam filters for years can potentially still be a scam. Fraud-
while SHAKEN is only now emerging sters can often gain access to fully ver-
to help identify illegitimate voice calls. ified numbers for short windows of
Spam filters scan email before delivery time, and then vanish by the time any-
to compare the content against known one realizes they’re using those phone
scams. These filters are not perfect, but numbers. That’s why SHAKEN has also
they’re good enough to hold email spam been designed to simplify the call trace-
down to tolerable, if still slightly annoy- back process.
ing, levels. Traceback is exactly what it sounds
That’s not possible with a telephone like: It’s a process that begins with the
call—it’s not really feasible to disclose the person receiving the call, tracing the call
content of a call before it’s connected. back through carriers to the person or
SHAKEN does the next best thing, by organization that made the call. In the
making it possible to easily track calls United States, the United States Telecom
from the point where they physically Association currently leads an industry
enter the network (more on that shortly) traceback initiative to identify the ori-
and then establish a caller’s reputation. gin of illegal calls. Traceback is largely a
Reputation is determined in large part process of scanning call-detail records
by the level of attestation callers receive to correlate a call coming into carrier A
from carriers. Reputation is also deter- with a call going out from carrier B, and
mined by connecting callers to their then repeating the process for as many
orig-id, so that over time less-reputable carriers as necessary to reach the person
callers may be identified by the number or business that placed the call. The pro-
of complaints made about that caller. If cess is now semiautomated, but it’s still a
the carrier knows a call is originating on complicated, multistep process.
its own network and the caller has the SHAKEN simplifies traceback, turn-
right to use the number—and the car- ing it into a one-step process no matter
rier has not received complaints about how many carriers have been involved
that caller—then, generally speaking, in the call. The same digital signature
the carrier can be more confident the that authenticates a call’s orig-id and
caller is not a scammer. By being able attestation level identifies exactly where
to identify less-reputable calls as they a problem call entered the network. This
What Carriers Phone companies don’t always know
everything about a call. STIR/SHAKEN uses
Know levels of attestation so that carriers can
classify what they do know about each call.
A–ATTESTATION B–ATTESTATION C–ATTESTATION
Originates on carrier’s own Originates on carrier’s own Originates on some other
network network network
Carrier has confirmed who Carrier has confirmed Carrier has NOT confirmed
the caller is who the caller is who the caller is
Carrier has verified Carrier has NOT verified Carrier has NOT verified
caller’s right to use the caller’s right to use the caller’s right to use
phone number phone number the phone number
SPECTRUM.IEEE.ORG | DEC 2019 |  51
 method simplifies the process of tracing
illegal calls, and will enable authorities to
The World’s Best investigate many more complaints in the
same amount of time. In the United States,
ROBOTS GUIDE for example, enforcement is handled by
the Federal Trade Commission (FTC),
Is Here! the FCC, the FBI, and state and local law
enforcement. The agencies should have
an easier time coordinating their efforts
ROBOTS.IEEE.ORG with a simpler traceback tool.
It’s also possible that a less-legitimate
carrier could be tempted to solicit ille-
gal robocalls. After all, the carrier would
still be paid for the service by the caller.
Simpler tracebacks make it easier to spot
a pattern if, for instance, one carrier is
hosting a lot of illegal robocalls. While
mainstream carriers have no interest in
Sawyer Robot,
Courtesy of Rethink
hosting robocalls, SHAKEN removes the
Robotics, Inc. small temptation that fly-by-night carri-
ers might have to make money by solic-
IEEE Spectrum’s new iting these callers.
ROBOTS site features SHAKEN’s digital signatures also pro-
more than 200 robots vide hard evidence of the source of illegal
calls, making successful prosecution eas-
from around the world. ier. In June, the FTC announced that the
• Spin, swipe and tap to make agency had filed 145 cases to date against
robots move. illegal robocall operations. Of course,
• Read up-to-date robotics news. those 145 cases predate SHAKEN. It’s not
• Rate robots and check their ranking. a large number, although the FCC, for its
• View photography, videos part, did go up against some big players,
and technical specs. including one man, Adrian Abramovich,
• Play Faceoff, an interactive who made over 100 million robocalls and
question game. was fined US $120 million. SHAKEN won’t
stop robocalls directly, but it will be an
important tool in identifying, locating,
and prosecuting illegal robocallers and
those who support them at a far greater
rate. Given time, SHAKEN should have
a huge impact on how many robocalls
scammers can get away with, and how
many new actors attempt to start their
own scams.
That said, scammers are nothing if not
resourceful. They will find a new weak-
ness to exploit, just as they’ve exploited
a loophole for call spoofing. When weak-
nesses are discovered—and it is a question
of when, not if—SHAKEN must be adjusted
quickly to patch the vulnerability.
Check out Robots.ieee.org SHAKEN is being deployed in the United
States and Canada independently because
on your desktop, tablet,
the current specifications consider only
or phone now! how the protocol operates within a single
country. It’s our hope to extend it inter-
52  | DEC 2019 | SPECTRUM.IEEE.ORG
nationally. That’s important because,
as we’ve mentioned, scammers often
place robocalls internationally using the
Internet. The task force that developed
SHAKEN has already begun working on
expanding the protocol so that carriers
in one country can verify calls that have
been digitally signed in another country.
As robocalls are brought under control
in the United States and Canada, illegal
robocallers are likely to attack citizens
and businesses in other countries.
As you read this, calls are already
being signed and verified across live net-
works by major carriers in the United
States, with Canada following in 2020. So
you can now relax, knowing you’ll never
be bothered by a robocall again, right?
Unfortunately, SHAKEN can’t com-
pletely stop robocalls on its own. It’s a
tool that can be used by call-blocking
apps to reduce the number of unwanted
calls. It will also help differentiate
between legitimate calls and illegal calls,
so users will be less likely to be taken in
by scams. In addition, SHAKEN makes it
much faster and easier to find and sanc-
tion illegal callers.
And it bears repeating that when
SHAKEN begins to reduce illegal calls,
scammers won’t just give up. The indus-
try will need to be vigilant to understand
robocallers’ latest tricks for avoiding
SHAKEN and will need to regularly
adjust the way the protocol is used to
close the gap.
STIR/SHAKEN will make a difference.
Not overnight, but over time the num-
ber of illegal robocalls scammers place,
and the calls’ effectiveness, will decrease.
The user experience will be like that of
email spam. At one time, experts pre-
dicted that email would become useless
because no one would be able to find the
real email among all the spam. But the
industry deployed a variety of anti-spam
measures, and eventually the situation
improved. Email spam didn’t go away
(you still have a spam folder, after all),
but it has minimal impact. SHAKEN will
provide the first step for a similar assault
on unwanted robocalling. n
↗ POST YOUR COMMENTS at https://spectrum.ieee.org/
robocalls1219
 SNOTBOT CO NTI N U E D F RO M PAG E 4 5

MEASURING UP: The Morphometer image-analysis tool [left] compares one whale with others
of its body type, displaying the result as an image of the subject whale superimposed on a diagram
representing an average whale. The diagram is color coded—red at the center, blending to white at
the size of an average whale and to blue for larger whales. With the individual whale superimposed
on this color map, any red that shows around the whale means the whale is underweight; the more
underweight, the darker the red. Above, fluke-recognition software makes a match.

the head and tail and then measures the
whale’s length and width at each pixel
point along the outline of its body. Our
software tracks the altitude from which
the drone photographed the whale and
combines that data with camera speci-
fications entered by the drone opera-
tor, allowing the system to automatically
convert the measurements from pixels
to meters.
Morphometer compares this whale
with others of its body type, display-
ing the result as an image of the subject
whale superimposed on a whale-shape
color-coded diagram with zones indicat-
ing the average measurements of simi-
lar whales. It’s immediately obvious if
the whale is normal size, underweight,
or larger than average, as would be the
case with pregnant females [see illustra-
tion, “Measuring Up”].
For our early prototype, we input
parameters for a “normal” body shape
based on age, sex, and other factors. But
now Morphometer is in the process of
figuring out “normal” for itself by pro-
cessing large numbers of whale images.
Whale researchers who use their own
drones to collect whale photos have
been sending us their images. Eventu-
ally, we envision setting up a collabor-
ative website that would allow images
and morphometry models to be shared
among researchers. We also plan to
adapt M ­ orphometer to analyze videos
INTEL (2)
frames or clips in which the whale’s posi-
tion and visibility are the best.
To help researchers gain a more com-
plete picture, we’re building statistical
models of various whale populations,
which we will compare to models derived
from human-estimated measurements.
Then we’ll take new photos of whales
whose age and gender are known, and
see whether the software correctly
classifies them and gives appropriate
indications of health; we’ll have whale
biologists verify the results.
Once this model is working reliably,
we expect to be able to say how a given
whale’s size compares with those of its
peers of the same gender, in the same
region, at the same time of year. We’ll
also be able to identify historical trends—
for example, this whale is not skinnier
than average compared with last year,
but it is much skinnier than whales in
its class a few decades ago, assuming
comparison data exists. If, in addition,
we have snot from the same whale, we
can create a more complete profile of
the whale, in the same way your credit
card company can tell a lot about you
by integrating your personal data with
the averages and variances in the gen-
eral population.
So far, SnotBot has told us a lot about
the health of individual whales. Soon,
researchers will start using this data to
are known as “apex predators,” mean-
ing they are at the top of the food chain.
Humpback whales in particular are gen-
eralist foragers and have wide-ranging
migration patterns, which make them
an excellent early-warning system for
environmental threats to the ocean as
a whole.
This is where SnotBot can really
make a difference. We all depend on
the oceans for our survival. Besides the
vast amount of food they produce, we
depend on them for the air we breathe:
Most of the oxygen in the atmosphere
comes from marine organisms such as
phytoplankton and algae.
Lately, ocean productivity associated
with a North Pacific warm-water anomaly,
or “blob,” has resulted in a reduction of
births and more reports of skinny whales,
and that should worry us. If conditions
are bad for whales, they’re also bad for
humans. Thanks to Project ­SnotBot, we’ll
be able to find out—accurately, efficiently,
and at a reasonable cost—just how the
health and numbers of whales in our
oceans are trending. With that informa-
tion, we hope, we will be able to spur
society to take steps to protect the oceans
before it’s too late. n
The whale images in this article were
obtained under National Marine F
­ isheries
Service permits 18636-01 and 19703.

↗ POST YOUR COMMENTS at https://spectrum.ieee.org/

of whales, automatically extracting the monitor the health of oceans. Whales snotbot1219

SPECTRUM.IEEE.ORG | DEC 2019 |  53
 PAST FORWARD_BY ALLISON MARSH

A GAME Cavity Sam, the cartoon patient in the board game Operation, suffers from an array of anatomically
THE STRONG, ROCHESTER, N.Y.

questionable ailments: writer’s cramp (represented by a tiny plastic pencil), water on the knee (a bucket of water),
WITH BUZZ butterflies in the stomach (you get the idea). Each player takes a turn as Sam’s doctor, using a pair of tweezers
to try to remove the offending piece. Dexterity is key. If the tweezers touch the side of the opening, it closes
a circuit, causing the red bulb that is Sam’s nose to light up and a buzzer to sound. Your turn is then over. The
game’s main flaw, at least from the patient’s perspective, is that it’s more fun to lose than to play perfectly. ■
↗ For more on electrified board games, see https://spectrum.ieee.org/pastforward1219

54  | DEC 2019 | SPECTRUM.IEEE.ORG
 Faculty Positions
Baylor University is a private Christian university and a nationally ranked
research institution, consistently listed with highest honors among
The Chronicle of Higher Education’s “Great Colleges to Work For.”
The University is recruiting new faculty with a deep commitment to
excellence in teaching, research and scholarship. Baylor seeks faculty
who share in our aspiration to become a tier one research institution
while strengthening our distinctive Christian mission as described in our
strategic vision, Pro Futuris, (www.baylor.edu/profuturis/) and academic
strategic plan, Illuminate (baylor.edu/illuminate). As the world’s largest
Baptist University, Baylor offers over 40 doctoral programs and has more
than 17,000 students from all 50 states and more than 85 countries.
Baylor seeks to fill the following two faculty position within the Electrical
and Computer Engineering Department: (i) Tenured/Tenure-Track
Assistant/Associate/Full Professor and (ii) Clinical Assistant/Associate
Professor. For tenured/tenure-track faculty, applicants must possess an
earned doctorate and an active research agenda in the area of wireless
and microwave circuits and systems. For clinical faculty, applicants must
possess an earned masters or doctorate degree, extensive industry
experience, and demonstrate the potential for excellent teaching. The
ECE department offers B.S., M.S., M.E., and Ph.D. degrees and is
rapidly expanding its faculty size, with current focus areas in wireless
& microwave, biomedical, computer, cyber-physical, materials, and
sustainable energy. Facilities include the Baylor Research and Innovation
Collaborative (BRIC), a newly-established research park minutes from
the main campus.
Applications will be considered on a rolling basis until the January 1,
2020 deadline. Applications must include:
1) a letter of interest that identifies the applicant’s anticipated rank,
2) a complete CV,
3) a concise statement of research agenda (tenure track only),
4) a concise statement of teaching interests, and
5) the names and contact information for at least four professional references.
Additional information is available at www.ecs.baylor.edu. Should
you have any questions on the position, feel free to contact the search
chair, Dr. Keith Schubert at keith_schubert@baylor.edu. Materials will be
submitted through Interfolio. For a direct link to the application portal,
please go to jobs.baylor.edu
Baylor University is a private not-for-profit university affiliated with
the Baptist General Convention of Texas. As an Affirmative Action/
Equal Opportunity employer, Baylor is committed to compliance with
all applicable anti-discrimination laws, including those regarding age,
race, color, sex, national origin, marital status, pregnancy status, military
service, genetic information, and disability. As a religious educational
institution, Baylor is lawfully permitted to consider an applicant’s
religion as a selection criterion. Baylor encourages candidates of the
Christian faith who are women, minorities, veterans, and individuals with
disabilities to apply.
Carnegie Mellon University Africa (CMU-Africa) invites
applications for teaching-track faculty positions.
The College of Engineering at Carnegie Mellon University, a
world leader in information technology, engineering and
artificial intelligence, has extended its global reach into
Africa. Offering master’s degrees to full-time resident
students mainly from across Africa at our base in Kigali,
Rwanda, CMU-Africa is educating future leaders who will
create the technology and business innovations that will
transform Africa.
We are seeking highly-qualified candidates at all levels to
join our dynamic, world-class faculty in contributing to the
emerging knowledge-based economies across Africa. Our
faculty members collaborate with industry to deliver
innovative, interdisciplinary graduate teaching and research
programs in the African setting. We are particularly
interested in applicants with expertise in software
engineering, learning science and education technology,
cybersecurity and privacy, data science and machine
learning, and artificial intelligence.
Teaching-track faculty focus mostly on strengthening our
graduate education and mentoring mission but are
encouraged to conduct research as well.
Candidates should possess a Ph.D. from a leading research
university and have a serious interest in teaching in the
context of opportunities in Africa. Applicants who have
passion for a culturally diverse environment and who
demonstrate a willingness to nurture the inclusive Carnegie
Mellon environment are encouraged to apply.
We take pride and active steps in considering a diverse
applicant pool in terms of gender, race, veteran status, and
disability. Carnegie Mellon University further seeks to meet
the needs of dual-career couples.
We will consider applications fitting our needs throughout
the academic year. Carnegie Mellon is an EEO/Affirmative
Action Employer -- M/F/Disability/Veteran.
Apply here:
www.africa.engineering.cmu.edu/faculty-positions
SPECTRUM.IEEE.ORG | DEC 2019 | 55
 TENURE-TRACK AND TENURED POSITIONS
ShanghaiTech University invites highly
qualified candidates to fill multiple tenure-
track/tenured faculty positions as its core
founding team in the School of Information Science and Technology (SIST).
We seek candidates with exceptional academic records or demonstrated strong
potentials in all cutting-edge research areas of information science and technology.
They must be fluent in English. English-based overseas academic training or
background is highly desired.
ShanghaiTech is founded as a world-class research university for training
future generations of scientists, entrepreneurs, and technical leaders. Boasting
a new modern campus in Zhangjiang Hightech Park of cosmopolitan Shanghai,
ShanghaiTech shall trail-blaze a new education system in China. Besides establishing
and maintaining a world-class research profile, faculty candidates are also expected
to contribute substantially to both graduate and undergraduate educations.
Academic Disciplines:
Candidates in all areas of information science and technology shall be considered.
Compensation and Benefits:
Salary and startup funds are highly competitive, commensurate with experience
and academic accomplishment. We also offer a comprehensive benefit package
to employees and eligible dependents, including on-campus housing. All regular
ShanghaiTech faculty members will join its new tenure-track system in accordance
with international practice for progress evaluation and promotion.
Qualifications:
• Strong research productivity and demonstrated potentials;
• Ph.D. (Electrical Engineering, Computer Engineering, Computer Science,
Artificial Intelligence, Financial Engineering, Signal Processing, Operation
Research, Applied Math, Statistics or related field);
• A minimum relevant (including PhD) research experience of 4 years.
Applications:
Submit (in English, PDF version) a cover letter, a 2-page research plan, a CV plus
copies of 3 most significant publications, and names of three referees to: sist@
shanghaitech.edu.cn. For more information, visit http://sist.shanghaitech.edu.
cn/2017/0426/c2865a23763/page.htm
Deadline: The positions will be open until they are filled by appropriate candidates.

Columbus, Ohio
Open Tenure-Track Faculty Position
Department of Electrical and Computer Engineering
ElectroScience Laboratory

The Ohio State University and the
Department of Electrical and Computer
Engineering invite applications for a
tenure track faculty position in applied
electromagnetics to support priorities
of the ElectroScience Laboratory (ESL), a
major center of excellence in the College of
Engineering. We are especially interested
in applicants with research interests in
photonics, bio-electromagnetics, THz and
millimeter-wave circuits, data analytics
and machine learning in electromagnetics,
mixed-signal microelectronics,
radio frequency systems (radar,
communications, and navigation), remote
sensing, antennas and propagation, and
quantum technologies. While the search
is focused on assistant professor positions,
outstanding candidates at the associate
level will also be considered.
Required Qualifications: Applicants
must have a Ph.D. and outstanding
academic credentials. Successful
candidates are expected to develop a
vigorous externally funded research
program, show excellence and leadership
in academic and scholarly activities, and
demonstrate outstanding teaching at the
undergraduate and graduate levels.

How to Apply
Submit applications electronically at https://academicjobsonline.org/ajo/jobs/15118
Application materials must include a cover letter, curriculum vitae,
statements of research and teaching interests, contact information for three
references, and three authored papers.
The Ohio State University College of Engineering is strongly committed to promoting diversity and
inclusion in all areas including scholarship, instruction and outreach. In the cover letter, please
describe experiences, current interests or activities, and/or future goals that promote a climate
that values diversity and inclusion in one or more of these areas.
We will begin reviewing applications on December 1, 2019 and continue until the position is filled.

56 | DEC 2019 | SPECTRUM.IEEE.ORG
 Electrical and Computer
Engineering, Duke University
Tenure-track position
The Electrical and Computer Engineering Department
at Duke University invites applications for a tenure-
track faculty position to begin in Fall 2020. The
position is primarily at the junior level, and graduating
students, postdocs, or junior faculty are encouraged
to apply. Candidates must have a well-demonstrated
background, PhD or equivalent degree in ECE or
related area, and contributions both in AI (in its
broader definition) and health, and their combination.
This position is part of the broad initiative at Duke
University and Duke Health, with the goal to make
our leadership in AI, health, and their intersection,
even stronger. Successful candidates should
be committed to high quality scholarly activities
including the development of a vibrant research
program supported by external funding, and active
participation in teaching at both the undergraduate
and graduate levels. Applications received prior
to December 15, 2019 will receive priority in
consideration, see https://academicjobsonline.org/
ajo/jobs/14926
Duke is an Affirmative Action / Equal Opportunity Employer
committed to providing employment opportunity without
regard to an individual’s age, color, disability, gender, gender
expression, gender identity, genetic information, national
origin, race, religion, sex, sexual orientation, or veteran status.
The ECE Department (https://engineering.wayne.edu/
ece/) at Wayne State University invites applications
for two faculty positions in computer architecture,
computer systems, cloud computing, and edge
computing. Applications must be submitted at http://
jobs.wayne.edu, posting #043981 or #044760, based
on research areas. The successful candidates must
have the ability to develop and maintain a productive
research program, including external funding,
supervise the thesis and dissertation research of
graduate students, teach undergraduate and graduate
courses and contribute to college and university service
activities. Candidates should possess Ph.D. degrees
in Electrical and Computer Engineering or a closely
related field. We expect to make the appointment at the
Assistant Professor level, but will consider exceptional
candidates at the Associate Professor level. Wayne
State University is a premier, public, urban research
university located in the heart of Detroit where
students from all backgrounds are offered a rich,
high quality education. Our deep‐rooted commitment
to excellence, collaboration, integrity, diversity and
inclusion creates exceptional educational opportunities
preparing students for success in a diverse, global
society. WSU encourages applications from women,
people of color, and other underrepresented people.
Wayne State is an affirmative action/equal opportunity
employer. Applications submitted before January 17,
2020 will receive full consideration. The positions will
remain open until filled.
Tenure Track Faculty Positions in
Electrical, Computer, & Biomedical Engineering
Located in the heart of downtown Toronto, the largest and most culturally diverse city in Canada,
Ryerson University is known for innovative programs built on the integration of theoretical and
practical learning. Our undergraduate and graduate programs are distinguished by a professionally
focused curriculum with a strong emphasis on excellence in teaching, scholarly, research and
creative activities.
Position Requirements
The Department of Electrical, Computer, & Biomedical Engineering in the Faculty of Engineering and
Architectural Science at Ryerson University, invites applications for THREE (3) full-time tenure-track
positions, at the Assistant Professor level, with a start date of July 1, 2020, subject to final budgetary
approval. Candidates must have a Ph.D. degree by the time of appointment in Electrical, Computer or
Biomedical Engineering, or in related discipline. Positions are as follows:
1) Faculty Position in Electrical Engineering specializing in Artificial Intelligence & Machine Learning
The ideal candidate must be able to demonstrate sound expertise and specialization in artificial
intelligence and machine learning, and their application to areas such as computer vision, signal
processing (audio, image, video, etc.), consumer electronics, cyber security, energy and power
systems.
2) Faculty Position in Electrical Engineering specializing in Microelectronic Design
The ideal candidate must be able to demonstrate sound expertise in one or more of the following areas
of design specialization: (i) ultra-low-power integrated circuits for application such as embedded
sensors, implanted devices, Internet of Things; (ii) high-speed data communication for applications
such as cloud-based computing, computer servers, multi-core computer systems; (iii) wireless data
communications targeting wireless sensors and sensor networks, cellular networks; (iv) AI-inspired/
enabled processors.
3) Faculty Position in Biomedical Engineering
The ideal candidate must be able to demonstrate sound expertise in one or more of following areas:
BioMEMS, Bio-Robotics, Bioinformatics, Medical Devices or Physiological Modeling
In addition, the following are expectations for all candidates in the above-mentioned positions:
• Demonstrate strong experience in undergraduate course development and teaching. Ability to
effectively teach key fundamental and applied engineering courses;
• Strong research profile with evidence of peer reviewed publications/contributions and external
grants of an internationally competitive caliber;
• Demonstrated ability to establish and maintain an independent, externally funded research program;
• Evidence of research collaborations or strong potential/ability to attract industrial collaborative
initiatives;
Professional Engineering (P.Eng.) registration in the province of Ontario (or eligibility to register)
is a necessary condition for appointment. Please clearly indicate your status or eligibility. All
candidates must also have a demonstrated commitment to upholding the values of equity, diversity,
and inclusion as it pertains to service, teaching, and scholarly, research or creative activities.
HOW TO APPLY
All qualified candidates are invited to visit http://hire.ecb.ryerson.ca to apply by January 31, 2020.
The application must contain the following: a letter of application, a curriculum vitae, a statement of
research interests, teaching and research philosophies, results of teaching evaluations (or equivalent
evidence, such as a teaching dossier), and the names of at least 3 individuals who may be contacted
for reference letters. Please CLEARLY indicate in your application if you are a Canadian Citizen or a
permanent resident of Canada.
SPECTRUM.IEEE.ORG | DEC 2019 | 57
 Professor/Associate Professor/Assistant Professorship in
the Department of Electrical and Electronic Engineering

The Department of Electrical and Electronic Engineering, Southern University

of Science and Technology (SUSTech) now invites applications for the faculty
position in the Department of Electrical and Electronic Engineering. It is We seek dynamic tenure track Assistant
Professors in the general area of digital power
seeking to appoint a number of tenured or tenure track positions in all ranks. electronics and energy systems, but distinguished
Candidates with research interests in all mainstream fields of electrical and electronic engineering candidates at any rank and in other areas of
electrical engineering may be considered with
will be considered, including but not limited to IC Design, Embedded Systems, Internet of Things, commensurate qualifications, experience and
VR/AR, Signal and Information Processing, Control and Robotics, Big Data, AI, Communication/ record of excellence in teaching and research.
Networking, Microelectronics, and Photonics. These positions are full time posts. SUSTech adopts
Areas of special interest include real-time
the tenure track system, which offers the recruited faculty members a clearly defined career path. computing for electronic power control, design
Candidates should have demonstrated excellence in research and a strong commitment to teaching. tools and simulation methods for electronic
power systems, electronic power converters
A doctoral degree is required at the time of appointment. Candidates for senior positions must have in the grid, interfaces for distributed energy
an established record of research, and a track-record in securing external funding as PI. As a State- resources, dynamic interactions and stability in
level innovative city, it is home to some of China’s most successful high-tech companies, such electronic power systems. Strong potential for
as Huawei and Tencent. We also emphasize entrepreneurship in our department with good initial collaborating with other high-profile research
groups, such as the AI Institute, is preferred.
support. Candidates with entrepreneur experience is encouraged to apply as well.
All faculty are expected to contribute to our
To apply, please send curriculum vitae, description of research interests and statement on teaching research and educational missions and to
participate in service and outreach activities.
to eehire@sustech.edu.cn. SUSTech offers internationally competitive salaries, fringe benefits
including medical insurance, retirement and housing subsidy, which are among the best in China. We are committed to enhancing the diversity
Salary and rank will commensurate with qualifications and experience. of our faculty and student populations
and we particularly seek applicants who
More information can be found at http://talent.sustech.edu.cn/en and http://eee.sustech.edu.cn/ can help us increase the representation
en. The search will continue until the position is filled. of women and other minorities.
Apply at: https://uscjobs.sc.edu/postings/69141
For informal discussion about the above posts, please contact Chair Professor Xiao Wei SUN,
Head of Department of Electrical and Electronic Engineering, by phone 86-755-88018558 or email: Ph.D is required. Search remains open until filled.
sunxw@sustech.edu.cn. Review of applicants will begin on Jan. 1, 2020
The University is an equal employment
To learn more about working & living in China, please visit: http://www.jobs.ac.uk/careers- opportunity employer for veterans, individuals
advice/country-profiles/china. with disabilities and other protected categories.

Faculty Position in Electrical Engineering

Department of Electrical, Computer,
and Systems Engineering
Case Western Reserve University,
Cleveland, Ohio
Electrical and Computer Engineering The Department of Electrical, Computer, and
Systems Engineering at Case Western Reserve
Herbert Wertheim College of Engineering University (CWRU) invites applications for tenure-
University of Florida track faculty positions in Electrical Engineering.
Preference will be given to the Assistant Professor
level, but other ranks will also be considered, for
The Herbert Wertheim College of Engineering at the University of Florida invites applications for 9-month, ten- starting dates as early as July 1, 2020. Candidates
ure track, full-time, positions at the rank of Assistant Professor in the Department of Electrical and Computer must have a Ph.D. degree in Electrical Engineering
Engineering (ECE). The open positions are for candidates working in electronics, machine learning, or related or a related field.
application areas. Successful candidates for these positions must possess a PhD in an engineering-related field The search is in the broader areas of micro/
nanosystems and integrated circuits, with a
and show potential for developing a successful research and teaching program.
strong emphasis in applications related to human
The University of Florida is the flagship campus of the State of Florida university system and, according to US health and symbiotic integration of humans with
machines in wearable and implantable fashion.
News and World Report, ranks 7th among best public universities in the US. The University of Florida has 16 In micro/nanosystems, the search emphasizes
research active colleges including a College of Medicine, Veterinary Science, Journalism, Food Science and Agri- expertise in novel devices, heterogeneous
culture providing an excellent environment to build highly trans-disciplinary research programs. The ECE Depart- integration, flexible/wearable systems, and
advanced packaging. In circuits/instrumentation,
ment offers BS, MS and PhD degree programs with an enrollment of about 1200 full-time undergraduate students the department invites candidates with expertise
and 600 graduate students of which about 200 are PhD students. Currently, the ECE Department has 54 tenured/ in analog/mixed-signal integrated circuits for
tenure-track faculty which include 17 IEEE fellows, 17 NSF CAREER Award winners, and 7 PECASE winners. The sensor interfacing. The department is particularly
Department’s external research expenditures were $19M last year (ASEE definition). The ECE graduate program interested in candidates with experience in both
focus areas. Outstanding candidates in micro/
ranks 17th among public ECE departments according to U.S. News and World Report. nanosystems and integrated circuits in applications
related to energy, advanced manufacturing and
The search committee will begin reviewing applications immediately and will continue to receive applications until aerospace will also be strongly considered.
the position is filled. Faculty position applications are processed through Interfolio, UF’s partner in faculty hiring. Additional information about the position,
Details and application information are available at: https://apply.interfolio.com/54353. Applicants will need to department, and application package is available at
create an Interfolio account using either the single sign-on option, available at the “partner institution” button, or http://engineering.case.edu/eecs/jobs.
with a personal email address. The application will require the following documents: curriculum vitae, statement CWRU provides reasonable accommodations to
of research interests, statement of teaching philosophy, and contact information for three references. applicants with disabilities. Applicants requiring an
accommodation for any part of the application and
The University of Florida is an Equal Opportunity Employer. hiring process should call 216.368.8877.

58 | DEC 2019 | SPECTRUM.IEEE.ORG
 Professor/ Associate Professor/ Assistant Professor
School of Microelectronics Tenure-Track Assistant/Associate Professor Positions
Southern University of Science and Department of Electrical Engineering
Technology (Shenzhen, China) University of North Texas
Post Specifications
School of Microelectronics (SME) National Exemplary School of Microelectronics, The Department of Electrical Engineering at the University of North
Southern University of Science and Technology (SUSTech) invites highly qualified
candidates to fill multiple tenure-track/tenured faculty positions in the areas of Texas (UNT) is seeking candidates for multiple faculty positions at the
(but not limited) Emerging Microelectronic Devices (Wide-bandgap, Nonvolatile Assistant (tenure-track) or Associate Professor (tenure-track or tenured)
memory, MEMS Sensor), and IC-Chip Designs (Future Computing/Communication/
Biomedical SoC). level starting Fall 2020. The positions are open to all areas in electrical
Junior applicants should have (i) a PhD degree in related fields; and (ii) outstanding engineering with an emphasis on RF, electromagnetics and electronic
potential in teaching and research. Candidates for senior post are expected to have
demonstrated exceptional academic leadership and strong commitment to be materials, analog and digital circuit design, embedded systems, control
excellent in teaching, research, and services.
and power systems. An earned Ph.D. degree in electrical engineering or
Applications
Submit (in English, PDF version) a cover letter, a statement in research and
a closely related field is required. For the Assistant Professor positions,
teaching, a CV plus copies of 3 most significant publications, and contacts of three we are looking for candidates with a strong publication record and
referees to: sme-hr@sme.sustech.edu.cn entitled with “Apply for Faculty Position”.
Applicants are required to specify the rank of the position in their letter of the potential to succeed in securing research funding and mentoring
application. The positions will be open until they are filled by appropriate candidates.
graduate students. For the Associate Professor position, we expect a
For more information, please visit http://ohr.sustc.edu.cn/sustczp/product/
recruit/a.do?action=toZPGWList2&entityId=T_RECRUIT_PLAN&postType=2&se sustained record of publications, mentoring junior faculty, advising
lectedId=1009881. graduate students, service to the University and profession, and securing
Salary and Fringe Benefits external funding for research activities.
Salary and startup funds are highly competitive, commensurate with experience
and academic accomplishment. All regular faculty members will join the tenure-
track system in accordance with international practice for progress evaluation and All applications must be submitted online at: http://facultyjobs.unt.edu.
promotion. Applicants are encouraged to check out the details about the university The committee will begin its review of applications on November 30, 2019,
at: http://www.sustech.edu.cn.
and continue to accept and review applications until the positions are filled.

The Electrical Engineering Department offers BS, MS, and Ph.D. degrees
in electrical engineering. It is home to over 440 undergraduate and
graduate students. Additional information about the department is
available at the website: http://engineering.unt.edu/electrical/.
Department of Electrical and Computer Engineering
University of North Texas (UNT) is a “R1 Doctoral University with the
Graduate School of Engineering and Management
highest research activity”. UNT is located in Dallas-Fort Worth, one of
Air Force Institute of Technology (AFIT)
the fastest growing metropolitan areas with a population of about 7
Dayton, Ohio
million people, an ever-increasing industrial and business activities, and
Faculty Position
a reasonable cost of living. The vibrant UNT College of Engineering has
The Department of Electrical and Computer Engineering at the Air Force
Institute of Technology is seeking applications for a tenured or tenure- more than 100 faculty members. The college’s faculty also boast two
track faculty position. All academic ranks will be considered. Applicants National Academy memberships, more than 20 faculty fellowships across
must have an earned doctorate in Electrical Engineering, Computer
Engineering, Computer Science, or a closely affiliated discipline by the more than 15 societies, and five NSF CAREER Awards. The college is
time of their appointment (anticipated 1 September 2020). equipped with numerous state-of-the-art facilities and excels in the areas
We are particularly interested in applicants specializing in one or more of
the following areas: autonomy, artificial intelligence / machine learning,
of artificial intelligence, autonomous vehicles, digital manufacturing,
navigation with or without GPS, cyber security, and VLSI. Candidates in health sciences, and sensor systems with a recent $10 million boost to
other areas of specialization are also encouraged to apply. This position
requires teaching at the graduate level as well as establishing and the Center for Agile and Adaptive Additive Manufacturing. The college
sustaining a strong DoD relevant externally funded research program with plans to hire 17 tenure and tenure track faculty this year and to expand
a sustainable record of related peer-reviewed publications.
the college faculty by 50% in the next 5 years.
The Air Force Institute of Technology (AFIT) is the premier Department of
Defense (DoD) institution for graduate education in science, technology,
engineering, and management, and has a Carnegie Classification as a The University of North Texas System and its component institutions are committed
High Research Activity Doctoral University. The Department of Electrical to equal opportunity and comply with all applicable federal and state laws regarding
and Computer Engineering offers accredited M.S. and Ph.D. degree
programs in Electrical Engineering, Computer Engineering, and Computer nondiscrimination and affirmative action. The University of North Texas System
Science as well as an MS degree program in Cyber Operations.
and its component institutions do not discriminate on the basis of race, color, sex,
Applicants must be U.S. citizens. Full details on the position, the
department, applicant qualifications, and application procedures can be sexual orientation, gender identity, gender expression, religion, national origin, age,
found at http://www.afit.edu/ENG/ . Review of applications will begin disability, genetic information, or veteran status in its application and admission
on January 6, 2020. The United States Air Force is an equal opportunity,
affirmative action employer. processes, educational programs and activities, and employment practices.

SPECTRUM.IEEE.ORG | DEC 2019 | 59
 Multiple Tenure Track Positions in
Electrical and Computer Engineering
The ECE Department at the University of Michigan-Dearborn invites
applications for three tenure-track Assistant Professor positions. We
are seeking qualified individuals in the following areas:
Power & Energy, especially as it relates to smart grid, cyber-physical
system, power electronics, and motor driver for electric vehicles.
Robotics, all areas, relevance to manipulation, reinforcement learning,
mobile and humanoid robots, and autonomous vehicles is desirable.
Computer Engineering and Embedded Systems, all areas, Computing
& Networks, Embedded Systems, Machine Learning, Cybersecurity,
and Intelligent Systems.
The expected starting date is September 1, 2020. Consideration is
primarily at the Assistant Professor level, but outstanding candidates
at the Associate Professor level will be considered. Applicants should
have research interests and expertise within the areas listed above.
Candidates must have earned a Ph.D. degree in computer science or
a closely related discipline by September 1, 2020. Candidates will be
expected to do scholarly and sponsored research, as well as teaching
at both the undergraduate and graduate levels.
Qualified applicants are invited to send a cover letter, curriculum
vitae, statements of teaching and research, and a list of three to five
references through Interfolio at:
Power/Energy Systems: http://apply.interfolio.com/68775
Robotics Engineering: http://apply.interfolio.com/68774
Computer Engineering: http://apply.interfolio.com/68776
Tenure Track Faculty Position
Review of applications will begin immediately and continue until a The UNC Charlotte Department of Electrical and Computer Engineering
suitable candidate is appointed to each of the areas above. However, (ECE) invites applications for a tenure-track Assistant or Associate
Professor starting in Fall 2020. Candidates are sought in control
applications received prior to January 15, 2020 will receive highest priority. systems with interest in linear/non-linear control theory with application
The University of Michigan-Dearborn is one of three campuses of the to robotics, autonomous systems, intelligent systems, and embedded
systems. Exceptional candidates in related areas are encouraged to apply.
University of Michigan. The University offers high quality undergraduate,
Candidates should be committed to excellence in teaching at the
graduate, professional and continuing education to residents of undergraduate and graduate levels, development of sponsored
Southeastern Michigan. Our faculty come from respected doctoral research programs, supervision of student research, and academic
programs and recognized for excellence in research and teaching. services. Priority will be given to candidates having original and
promising research as demonstrated by peer-reviewed publications,
The campus is located on 200 acres of the original Henry Ford Estate, potential for success in extramural funding, and a plan for future
centrally located within one of America’s largest business regions. The research that integrates well with existing department strengths.
A Ph.D. degree in Electrical or Computer Engineering or equivalent
geographically diverse area provides faculty with a variety of urban, is required for candidates hired at the Assistant Professor or the
suburban, and rural areas within a reasonable commute, including Associate Professor level. Hires at the rank of an Associate Professor
Detroit, Detroit suburbs, and Ann Arbor. are also expected to have demonstrated success in receiving
extramural research funding, as well as demonstrated success in the
At University of Michigan-Dearborn, a university that provides progression of graduate students.
the academic excellence of the University of Michigan in service The ECE department offers B.S. degrees in both Electrical Engineering
of southeast Michigan. We are an inclusive, student-focused and Computer Engineering, and M.S. and Ph.D. degrees, with
approximately 700 undergraduate and 200 graduate students; see
institution, committed to excellence in teaching, learning, research http://ece.uncc.edu.
and scholarship, as well as access, affordability and metropolitan Applications must be made electronically at https://jobs.uncc.edu
impact. UMD is rich in opportunities for independent and collaborative (Position # 004395). Applicants should submit a cover letter, a complete
research, integrated learning and civic engagement. We are responsive resume, a statement of teaching and research experience/goals, and
contact information of at least three references. Review of applications
to the changing needs of our diverse student body, the world in which will begin immediately, and the position will remain open until filled.
they live and work, ever-advancing technologies, and a knowledge- Women and minority candidates are encouraged to apply. Candidates
based global economy. having demonstrated ability to contribute to diversity initiatives are
encouraged to apply.
The University of Michigan-Dearborn is an equal opportunity/ The University of North Carolina at Charlotte is
affirmative action employer. an EOE/AA employer and an ADVANCE Institution.

60 | DEC 2019 | SPECTRUM.IEEE.ORG
 Come work with us!

The Massachusetts Institute of Technology (MIT) Department of Mechanical Engineering together with the Institute for Data, Systems, and Society
(IDSS) seeks candidates for tenure-track faculty positions in Learning for Dynamics and Control, Autonomy, and Intelligent Systems to start July 1,
2020 or on a mutually agreed date thereafter. Appointments will be at the assistant or untenured associate professor level. In special cases, a senior
faculty appointment will be considered.
We are looking for candidates with proven excellence in research who have the vision and interest to contribute to interdisciplinary research in
autonomy and robotics, with fundamental expertise in learning for dynamics and control theories, nonlinear dynamics and physics-based modeling,
uncertainty quantification, high dimensional statistics, science of autonomy, intelligent systems, and/or data-driven science and engineering.
We seek candidates who can go beyond current approaches to machine learning in computer vision and natural language processing and address
the explosion of real-time data that is emerging from the physical world. Dealing with challenges in data-enabled control and decision making for
physical and embedded systems requires a rapprochement of areas such as machine learning, dynamical systems, decision theory, and model-
based control with domains such as intelligent autonomous systems, transportation and logistics, power networks, smart cities, manufacturing,
healthcare, and smart services.
The Department of Mechanical Engineering and IDSS are committed to fostering interdisciplinary research that can address grand challenges facing
our society. We seek candidates who will provide inspiration and leadership in research, contribute proactively to both undergraduate and graduate
level teaching in the Mechanical Engineering department and IDSS and add to the diversity of the academic community.
Faculty duties include teaching at the undergraduate and graduate levels, advising students, conducting original scholarly research and developing
course materials at the undergraduate and graduate levels. Candidates must hold an earned Ph.D. in a field related to Engineering, Physics, Data
Science, Computer Science, or Applied Mathematics or a related field by the beginning of employment.
In addition to this search, the Mechanical Engineering department has positions available broadly in mechanical engineering:
http://meche.mit.edu/faculty-positions
Applicants should send a curriculum vitae, a research statement, a teaching statement, and copies of no more than three publications.
They should also arrange for four individuals to submit letters of recommendation on their behalf. This information must be entered
electronically at the following site: https://school-of-engineering-faculty-search.mit.edu/meche-idss/ by December 15, 2019 when
review of applications will begin.
MIT is an equal-opportunity/affirmative action employer. Women and underrepresented minorities are
especially encouraged to apply. http://web.mit.edu

Professor (Open Rank)

The Department of Electrical and Computer Engineering at the University
of Maryland, College Park, invites applications for multiple tenure-track
faculty positions. Applicants should have a demonstrated record of research
achievements and publications in one of the following areas:
a) Machine learning algorithms and applications with methodologies from
signal processing, computer vision, and/or optimization;
b) Internet of Things (IoT) broadly defined, including hardware/software
methodologies of scalable, secure and resilient infrastructure for connected
smart devices, data analytics, edge and cloud computing; and
c) Quantum information processing and technology, particularly with a focus
on applications at the intersection of engineering and physical science.
A Ph.D. inIEEE
Electrical/Computer Engineering or a related discipline is required.
Publication: Size:
Candidates should be creative and adaptable and should have a high potential
Job# 54779 IO#: 54779-54779
for both research and teaching.
Mechanical: dg, dg, mbb, mbb, mbb
For best consideration, applications should be submitted by 10 December 2019
online at https://ejobs.umd.edu (position number 125223). An application
should include a cover letter, curriculum vitae, a list of at least three references,
examples of research achievements including three significant publications, a
research statement (not exceeding three pages), and a statement of teaching
philosophy (not exceeding two pages). The applicant should clearly indicate in the
cover letter which of the above areas best fits the candidate’s research interests.
The University of Maryland is an Equal Employment/Affirmative Action employer. We do
not discriminate in hiring on the basis of sex, gender identity, sexual orientation, race,
color, religious creed, national origin, physical or mental disability, protected Veteran
status, or any other characteristic protected by federal, state, or local law.
Mechanical Science and Engineering
The Grainger College of Engineering
University of Illinois at Urbana-Champaign
Department of Mechanical Science and Engineering at the University
of Illinois at Urbana-Champaign invites applications for multiple full-
time faculty positions at all ranks. Senior and mid-career faculty are
encouraged to apply, but all qualified candidates will be considered.
Candidates are sought in all technical subdisciplines of mechanical
science and engineering including design, thermosciences, solid and fluid
mechanics, dynamics and control, energy applications, manufacturing,
and robotics and cyber-physical systems. Ideal candidates include those
who demonstrate evidence of a commitment to diversity, equity, and
inclusion through research, teaching, and/or service endeavors.
Qualified senior candidates may also be considered for tenured
Associate Professor and Full Professor positions as part of the Grainger
Engineering Breakthroughs Initiative.
A doctoral degree in Mechanical Engineering or a related field is
required at the start date, and salary will be commensurate with
qualifications and experience. To ensure full consideration, applications
7”must
x 4.75”
be received by November 28,Notes:
2019.color
Early applications are strongly
encouraged as interviews may take place during the application
Screen: period,
but a decision will not be made until after the closing date. Applications
Proofreader:
received after that date may be considered until positions are filled. The
expected start date is August 16, 2020, but other start dates may be
considered.
A full position description and information on how to apply can be found
on the University of Illinois at Urbana-Champaign online jobsite http://
jobs.illinois.edu. For further information regarding application procedures,
please address questions to: mechse-facultyrecruiting@illinois.edu.
The University of Illinois conducts criminal background checks on
all job candidates upon acceptance of a contingent offer.
The U of I is an EEO Employer/Vet/Disabled www.inclusiveillinois.illinois.edu

SPECTRUM.IEEE.ORG | DEC 2019 | 61
 Assistant, Associate or Full Professor, position number 85317, University
of Hawai‘i at Mãnoa (UHM), Department of Electrical Engineering, invites Best Ne
w
in
Journal
applications for a full-time, tenure-track faculty position, pending position clearance
15
STM 20

and availability of funds. To begin approximately August 1, 2020 or soon thereafter.

UHM, a Carnegie R1 research university, is a top-50 public university dedicated

to providing world-class teaching, research, and service. Collaboration, funding Become a
opportunities, resources, and research exposure may be found through the
department’s involvement and affiliation with the $5 million NSF CyberCorps Scholarship for Service
published author
Program (SFS); the NSA/DHS National Center of Academic Excellence in Cyber Defense Research (CAE-R); in 4 to 6 weeks.
the $25 million NSF Center for Science of Information (focusing on big data, information theory, and machine
Published online only, IEEE Access is ideal
learning); and Ike Wai, a $20 million NSF EPSCOR project for bio/nano sensing. Further information is for authors who want to quickly announce
available at http://ee.hawaii.edu. recent developments, methods, or new
products to a global audience.
We are seeking candidates with a strong research record in microelectronics and related areas. Of particular
interest are candidates in the areas of on-chip machine learning and biomedical electronics. However, • Submit multidisciplinary articles that
exceptional candidates in all areas are encouraged to apply. do not fit neatly in traditional journals
• Reach millions of
Duties: Teach and develop courses in electrical or computer engineering, develop an extramurally funded global users through
research program, publish outstanding work in leading scholarly journals, supervise graduate students, and the IEEE Xplore®
provide department, college, and university service. digital library with
free access to all
Minimum qualifications: An earned Ph.D. in Electrical or Computer Engineering, Computer Science, or a
closely related discipline, with a strong research track record. All-But-Dissertation cases will be considered
Included in
but dissertation must be filed before start of employment. For complete duties and qualifications, and Web of Science
application instructions, refer to: and has an
Impact Factor
https://www.governmentjobs.com/careers/hawaiiedu and search for Position 85317.

Continuous recruitment: Application reviews will begin on January 15, 2020, and will continue until
the position is filled. The University of Hawai‘i is an equal opportunity/affirmative action institution and

17-PUB-013 3/17
Learn more at:
encourages applications from women and minority candidates. ieeeaccess.ieee.org

The Department of Electrical and Computer Engineering at Oakland
University (OU) invites applicants for a tenure track position as
Assistant Professor in Electrical and Computer Engineering starting in
August 2020. We encourage applicants from various areas of electrical
and/or computer engineering to apply. Teaching, research, or hands-
on experience in intelligent controls and mechatronics, autonomous
systems, and/or electric transportation and drives is a plus.
Apply at: http://jobs.oakland.edu/postings/17913. Applicants must
have an earned Ph.D. degree in Electrical and Computer Engineering
or related fields. ABD will be considered. Research and teaching with
commitment to excellence is required. The successful candidates will
be expected to teach and develop undergraduate and graduate courses
and laboratories, initiate and integrate strong research programs,
direct graduate students, and attract external research funding.
OU is a Ph.D. offering public institution located in southeastern
Michigan at the intellectual center of the Automotive Industry, adjacent
to the Oakland Technology Park, and within the Oakland County
Automation Alley. OU offers an exemplary fringe benefit package. The
ECE Department has 20 faculty members and offers BSE, MS, and
Ph.D. programs.
For full consideration applications should be submitted by January
31, 2020. Review will continue until the position is filled. Oakland
University is an Affirmative Action/Equal Opportunity Employer and
encourages applications from women and minorities.
Assistant Professor and Lecturer Positions in Electrical
& Computer Engineering (ECE) at Rowan University
With dramatic increase in undergraduate and graduate enrollment, a new PhD
program and a new building allowing us to expand, our award winning and
nationally ranked ECE Department (www.rowan.edu/ece) is growing. To support
this growth, we invite applications from energetic and innovative teacher-scholars
for both tenure-track Assistant Professor and non-tenure track Lecturer positions.
For the tenure-track position, our emphasis is on computer engineering,
broadly including computer architecture, embedded systems, internet of things,
cyberphysical systems, cybersecurity and blockchain, low power design, hardware
/ architecture / computing security, high performance and distributed computing,
etc. Candidates with exceptional credentials and promising research agendas in
other areas may also be considered.
For Lecturer positions, our emphasis is on digital and embedded systems, computer
architecture, control systems, communication systems, electromagnetics and
mixed signal design.
Successful candidates for all positions will share our dedication to project-
based learning, will develop and teach innovative curriculum components at
both undergraduate and graduate levels, have excellent communication and
collaboration skills and genuine interest in student mentoring.
The successful candidate for the Assistant Professor position will, in addition
to the above, help grow our Ph.D. program, establish an externally funded and
sustainable research program, enhance the department’s research infrastructure,
conduct and publish high quality research, and supervise student research and
entrepreneurial efforts within Rowan’s hallmark Engineering Clinic sequence. For
full position description and instructions to apply, please use the following links:
Assistant Professor Position: https://go.rowan.edu/joinrowanece2020
Lecturer Position: http://go.rowan.edu/joinrowanecelecturer
Rowan University values diversity and is an equal opportunity employer. All positions are
contingent upon budget appropriations.

62 | DEC 2019 | SPECTRUM.IEEE.ORG

The School of Electrical and Computer Engineering
at Purdue University is seeking applications for
tenure-track positions at the Assistant or Associate
Professor level in Computer Engineering. We are
particularly interested in candidates in computer
systems and security. Successful candidates must
hold a Ph.D. degree in Electrical and Computer
Engineering, Computer Science, or a related
discipline. A background check will be required for
employment in this position. Submit applications
online https://tinyurl.com/purdue-ecesystems2019
Purdue University is an EOE/AA employer. All
individuals, including minorities, women, individuals
with disabilities, and veterans are encouraged to apply.
The IEEE Foundation
is raising awareness, creating
partnerships, and generating
financial resources needed to
combat global challenges.
Our goal is to raise $30
million by 2020.
DONATE NOW
ieeefoundation.org
Roy E. Wilkens Missouri Telecommunications Professorship
in Electrical and Computer Engineering
Missouri University of Science and Technology
Rolla, Missouri
https://ece.mst.edu/
The Department of Electrical and Computer Engineering (ECE) at the Missouri
University of Science and Technology (Missouri S&T) in Rolla, Missouri is seeking outstanding applicants for
the Wilkens Missouri Professorship in Electrical Engineering in the area of Telecommunications. The position is
available at the Full Professor or the Associate Professor rank for a rising star. The endowed position provides
unrestricted research funds. The salary is commensurate with the rank and qualifications of the candidate being
sought.
Successful candidates will be expected to have strong commitments to contributing to the departmental and
college research efforts, high-quality teaching at the undergraduate and graduate levels, service in the applicant’s
professional community and our institution, and increasing the diversity of the student body and faculty. Priority
will be given to candidates with a sustained record of scholarly research, attracting external funding, excellence in
teaching, and international prominence and leadership in areas related to telecommunications. Applicants must
hold a Ph.D. in Electrical Engineering or a closely related field.
Missouri S&T’s ECE Department has 32 faculty members, approximately 575 undergraduate and 160 graduate
students, and research expenditures of approximately $5.3M in the past fiscal year. Further details on the
department’s vision and strategic plan, activities, and research may be found at: https://ece.mst.edu/.
Interested candidates should electronically submit their application consisting of: a cover letter, current
curriculum vitae, research statement, teaching statement, diversity statement, and complete contact information
for at least four references to Missouri S&T’s Human Resources Office at: http://hr.mst.edu/careers/academic-
employment/ using Reference Number 32082. For full consideration, applicants must apply by January 3, 2020.
For more information, please contact the Search Committee Chair, Dr. Jag Sarangapani, at sarangap@mst.edu.
Missouri S&T is an AA/EEO employer and does not discriminate on the basis of race, color, national origin, ancestry, religion,
sex, pregnancy, sexual orientation, gender identity, gender expression, age, disability, protected veteran status, or any other
status protected by applicable state or federal law. Females, minorities, and persons with disabilities are encouraged to apply.
The university participates in E-Verify. For more information on E-Verify, please contact DHS at: 1-888-464-4218.
The George Washington University
Department of Electrical & Computer Engineering
Tenure Track Opening for Assistant/Associate
Professor in Autonomous Systems Hardware
The Department of Electrical and Computer Engineering (ECE) at the George
Washington University (GWU) invites applications for a tenured/tenure-track
faculty appointment starting as early as Fall 2020 at the rank of Assistant or Associate Professor in system design,
particularly in the area of digital hardware with applications to autonomous systems.
The School of Engineering and Applied Science is fundamentally committed to increasing the diversity of its
faculty and staff. We welcome nominations of and applications from members of underrepresented groups, and
others who would bring additional aspects of diversity to the university’s missions.
The successful candidate will complement existing research strengths in computer architecture, nanotechnology,
and communication networks, and create new synergies among mechanical engineering, biomedical engineering,
and computer science. Applicants with expertise in any aspect of autonomous systems hardware will be
considered, including but not limited to digital circuits and interfacing, digital chip design, and embedded and
reconfigurable hardware.
Applicants must have an earned doctorate in Electrical Engineering, Computer Engineering, or a relevant discipline
at time of appointment; must demonstrate a solid publication record; must have established or exhibit potential
to establish a strong, externally sponsored research program, commensurate with the rank they are seeking; and
must be committed to excellence in teaching at both undergraduate and graduate levels.
The ECE department has experienced significant research growth in the last few years. It is housed in a state-of-
the-art 500,000 sq. foot building in the heart of the nation’s capital. More information about the department is
available at http://www.ece.seas.gwu.edu/.
To apply, complete the online application at http://www.gwu.jobs/postings/71663/.
A cover letter indicating primary area(s) of expertise and desired rank, detailed CV, concise statements on
teaching and research, and full contact information for four professional references are required. Only complete
applications will be considered. Review of applications will begin on January 3, 2020 and will continue until
position is filled.
Employment offers are contingent on satisfactory outcome of standard background screening. The university
is an Equal Employment Opportunity/Affirmative Action Employer that does not unlawfully discriminate in any
of its programs or activities on the basis of race, color, religion, sex, national origin, age, disability, veteran
status, sexual orientation, gender identity or expression, or on any other basis prohibited by applicable law.
SPECTRUM.IEEE.ORG | DEC 2019 | 63
 THE IEEE APP:
Your mobile
University of Illinois at Urbana-Champaign gateway to
The Department of Electrical and Computer Engineering at the University of Illinois at Urbana-
Champaign (ECE ILLINOIS) invites applications for full-time faculty positions. Senior and mid-career
faculty are encouraged to apply but all qualified candidates will be considered at all levels and in
IEEE.
all areas of electrical and computer engineering, broadly defined, with particular emphasis in the
areas of Computational and Physical Aspects of Electronics and Electromagnetics; Power and Energy
Systems; Quantum Information Science; Bioacoustics and Bioimaging; Circuits - System on a Chip;
AI/Autonomous Systems; Robotics; Machine Vision; Embedded Computing Systems and the Internet
of Things; Data- Centric Computer Systems and Storage; Networked and Distributed Computing
Systems. Applications are encouraged from candidates whose degrees and research programs are in
core as well as broad interdisciplinary areas of electrical and computer engineering. Ideal candidates
include those who demonstrate evidence of a commitment to diversity, equity, and inclusion through
research, teaching, and/or service endeavors.
Of particular interest is quantum information science, including quantum computing, quantum
communication, quantum sensing, and the development of novel hardware and architectures for
quantum information processing. This position is part of a $15M investment in the new Illinois Quantum
Information Science and Technology (IQUIST) center, which also includes multiple hires in Physics,
Computer Science, and Mathematics.
Applicants for all positions at the assistant professor level must have an earned Ph.D. or equivalent degree,
excellent academic credentials, and an outstanding ability to teach effectively at both the graduate and
undergraduate levels. Successful candidates will be expected to initiate and carry out independent research
and to perform academic duties associated with our B.S., M.S., and Ph.D. programs.
Qualified senior candidates may also be considered for tenured Associate and Professor positions as part
of the Grainger Engineering Breakthroughs Initiative (GEBI), which is backed by a $100-million gift from
the Grainger Foundation.
ECE ILLINOIS is in a period of intense demand and growth, serving over 3000 students and averaging
7 new tenure-track faculty hires per year in recent years. Faculty in the department carry out research Download now
in a broad spectrum of areas and are supported by world-class interdisciplinary research facilities,
including the Coordinated Science Laboratory, the Information Trust Institute, the Parallel Computing
and get IEEE at
Institute, the Nick Holonyak Jr. Micro and Nanotechnology Laboratory, the Beckman Institute for
Advanced Science and Technology, the Carl R. Woese Institute for Genomic Biology, as well as
your fingertips.
several industrial centers and programs that foster international collaborations. The ECE Department
also supports and encourages faculty involvement with the Nation’s first engineering-based College
of Medicine that has opened on campus. The plans are to facilitate transition from engineering
breakthroughs into translational medical practices. The department has one of the very top programs
in the world, granting approximately 450 B.S. degrees, 100 M.S. degrees, 80 M.Eng. degrees, and 75
Ph.D. degrees annually. ECE ILLINOIS is housed in its new 235,000 sq. ft. building designed to set a
new standard in energy efficiency which is a major campus addition with minimal carbon footprint.
In order to ensure full consideration by the Search Committee, applications must be received by
December 1, 2019. Applications will be reviewed until suitable candidates are identified. Interviews
and offers may take place before the deadline but all applications received by the deadline would
receive full consideration. Salary will be commensurate with qualifications. Preferred starting date
is August 16, 2020, but is negotiable. Applications can be submitted by going to http://jobs.illinois.
edu and uploading a cover letter, CV, research statement, teaching statement, and statement on
commitment to diversity, along with names of three references. The statement on diversity should
address past and /or potential contributions to diversity, equity, and inclusion through research,
teaching, and/or service. For inquiry, please call 217-333-2302 or email ece-recruiting@illinois.edu.
The University of Illinois conducts criminal background checks on all job candidates upon acceptance
of a contingent offer.
The University of Illinois is an Equal Opportunity, Affirmative Action employer. Minorities, women,
veterans and individuals with disabilities are encouraged to apply. For more information, visit http://
go.illinois.edu/EEO. To learn more about the University’s commitment to diversity, please visit
https://engineering.illinois.edu/about/diversity.html.
We have an active and successful dual-career partner placement program and a strong commitment to work-life balance and
family-friendly programs for faculty and staff (http://provost.illinois.edu/faculty- affairs/work-life-balance/).

64 | DEC 2019 | SPECTRUM.IEEE.ORG
 Because
someone
looks up
to you.

Providing a secure financial future is in your hands.

IEEE Member Group 10-Year Level
Term Life Insurance.
To learn more*, visit IEEEinsurance.com/10Year
*For information on features, costs, eligibility, renewability, limitations and exclusions.
The Group 10-Year Level Term Life Insurance Plan is available only for residents of the U.S. (except territories), Puerto Rico and Canada (except
Quebec). This plan is underwritten by New York Life Insurance Company, 51 Madison Ave., New York, NY 10010 on Policy Form GMR. This plan is
administered by Mercer Consumer, a service of Mercer Health & Benefits Administration LLC. This coverage is available to residents of Canada
(except Quebec). Mercer (Canada) Limited, represented by its employees Nicole Swift, Pauline Tremblay and Suzanne Dominico, acts as a broker
with respect to residents of Canada.
Program Administered by Mercer Health & Benefits Administration LLC
In CA d/b/a Mercer Health & Benefits Insurance Services LLC
AR Insurance License #100102691 CA Insurance License #0G39709
85335 (12/19) Copyright 2019 Mercer LLC. All rights reserved.
MATLAB SPEAKS
DEEP
LEARNING
With just a few lines of MATLAB®
code, you can use CNNs and training
datasets to create models, visualize
layers, train on GPUs, and deploy to
production systems.

mathworks.com/deeplearning

©2019 The MathWorks, Inc.