Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Social Cardboard:

Pretotyping a Social Ethnodroid in the Wild


Janet Wiles, Peter Worthy, Kristyn Hensby, Marie Boden, Scott Heath, Paul Pounds,
Nikodem Rybak, Michael Smith, Jonathon Taufotofua, Jason Weigel,
The University of Queensland, Brisbane, Australia
j.wiles@uq.edu.au
Abstract—Pretotyping is a set of techniques, tools, and metrics
for gauging the interest in a product, prior to full-scale A. Ethnodroids: Robots as social actors and data collectors
development [1]. This late breaking report describes a Ethnodroid is a term we coined for research robots that
pretotyping case study of an ethnodroid – a robot that functions function as ethnographers. In social robotics they can provide
as an ethnographer – intended to engage with young children and an embodied presence for research into social interaction with
record their learning progress. The central requirement for the a target group.
project is that the robot will be able to interact socially with
children aged 1-6 years in tablet-based tasks. We developed a The OPAL project is part of a broad-ranging research
simple robot made of MDF (thick cardboard), added tablets for program to study technology and language. We are currently
the face and torso, and controlled a scripted interaction using developing a child-friendly ethnodroid called Opie which will
Wizard of Oz (WoZ). Children’s engagement with the robot was be used to evaluate social components of children’s
tested in an early learning centre which provided a relatively interactions with tablet-based learning tasks and games. The
structured environment (“in the lab”) and at a science fair which key requirements for Opie are that it be both child-safe and
provided a relatively unconstrained setting (“in the wild”). The robust in relatively unstructured settings as well as socially
rapid testing revealed distinct effects in the children’s attitudes engaging. Prior experience has shown that commercially
and behaviors in the two user contexts and provided insights into available robots such as the humanoid Nao [5] are not robust
form, sensors and analyses for the design process. enough to withstand a preschool environment unsupervised
over long periods.
Keywords— pretotyping; child-friendly robot; ethnodroid;
Wizard of Oz; testing robots in the wild; OPAL robot; cHRI The inspiration for our design is the child-sized robot,
RUBI which functioned unattended for a month in a preschool
I. INTRODUCTION [6]. We aim to retain RUBI’s safety, robustness and
engagement, but don’t need its range of functional
Pretotyping is an approach to product development components. Like RUBI, our robot aims to collect data on
systematized by Alberto Savoia at Google to explain why interactions within and between children, however, we also
products fail in the market despite being well designed [1]. In require external monitoring and aim to further develop an
design terminology, “prototyping” covers all aspects of associated portable lab-kit [7].
development (see for example [2, 3]), but in engineering, its
meaning is narrower and entails correct functioning. Like
functional prototyping, pretotyping involves building a scaled II. METHODOLOGY: IN THE LAB AND AT THE SCIENCE FAIR
down version of a product, but it serves a different purpose, Over a 6-week period, a range of robots were developed
and entails different methods. Functional prototyping addresses (see [8] for further details). The general form is a semi
questions related to building an envisioned product: “Can we anthropomorphic robot with stylized elliptical head, a torso
build it?”, “Will it work as expected?”, “How cheaply can we constructed of sheets of MDF secured to a wooden frame, and
build it?”, “How fast can we make it?” [1]. In contrast, arms constructed from foam tubes ending in paddle-shaped
pretotyping is a light-weight process for addressing questions hands. The frame and equipment behind the robot are covered
about what product to build: “Would people be interested in by a black curtain. A tablet in the head shows a face, and one
it?”, “Would they buy it if we build it?”, “Will they use it as in the torso is used for games (see Fig 1).
expected?”, “Will they continue to use it?” [1]. Pretotyping
short-cuts the expense of fully functional robot prototypes, Studies were conducted by a facilitator interacting with the
which inherently require working hardware and software and robot and children1. Social responses in the 100-200
entails substantial investment. millisecond range (which we designate “social moments”) are
important for fluid human-human social interaction, and the
In this study, we used Savoia’s pretotyping questions to facilitator filled this role when the robot was offline or slow to
organize insights gained from the different contexts of early respond. The Wizard was operated by the facilitator in the lab,
stage child-robot interaction tests. Pretotyping starts with and by an additional team member at the fair.
construction of a minimal system that need only enable pretend
interaction with the envisaged product. In our case, we needed In the lab, the robot was positioned in the corner of a small
children interacting with a robot and hence functions that will child-safe room. Participants were invited to interact with the
require effort for a fully functional prototype were delivered robot and a facilitator for 10 minutes. Testing was an ongoing
using Wizard of Oz (WoZ) [4]. process that enabled many small refinements of the design. In
the first month, 6 children aged 1-6y participated in the lab.

This research was funded by the ARC Centre of Excellence for the Dynamics of Language and APA scholarships to JWe and PW. We thank the OPAL team at
UQ, and the Temporal Dynamics of Learning Center and the RUBI project at UCSD for their input in the design process.
1
Ethical clearance: Title – “RUBI-Oz Design Pilot Test” clearance number: 15-PSYCH-PHD-42-JH

978-1-4673-8370-7/16/$31.00 © 2016 IEEE 531


At the science fair, the robot was of view and showed that sensor placement needed adjustment.
positioned in a roped-off corner The key finding for ethnography was the massive amounts of
of a large hall where young data generated from every trial. First-hand experience of
children and their carers could coding touch data resulted in rapid innovation in sensors to
see the robot prior to automate the touch analyses and visualizations to interpret the
participating. In one day (5 data [10][11], and prompted the design of automated analysis
hours of testing), 17 children tools much earlier than planned (an insight recommended in
between 1-6y engaged with the other studies [12]).
robot, in groups or alone (for
Q4. How long would it keep their attention? Savoia’s 4th
further details see [9][10]).
Fig 1. Opie at the fair. question concerns use over days to months. The study to date
has been limited to one-off interactions, and we have yet to
III. EVALUATION assess whether the robot will keep children’s attention over
Savoia’s pretotyping questions were adapted to children’s multiple visits. Tests over multiple days in schools are planned
interactions with the robot, with reflection on the process to address this question.
comparing and contrasting our expectations from prior
experience and the background literature. It should be noted IV. CONCLUSIONS: PRETOTYPING AN ETHNODROID
that our team comprised a wide range of disciplines (designers,
The key finding of the pretotyping with children is just how
developmental psychologists, engineers, cognitive scientists, as
little a robot needs to be social and engaging: A cardboard
well as students). Hence, there was no “team” expectation, frame, foam arms and two tablets can be socially engaging,
rather a range of different experiences and questions.
given the right context and Wizard. The social interactions are
Q1. Would children be interested in the robot? The tested not just between child and robot, but are structured by the
versions of the robot were only minimally viable. Few of our settings, the parents, facilitator, WoZ, and the robot’s
research team expected that it would engage children’s interest autonomous communication. For its ethnodroid function, it
for a substantial period of time. To our surprise, most children became obvious that the robot could generate more low level
were strongly attracted to the robot in both settings. In the sensor data than could be stored for analysis, and that
structured lab environment, all children were assigned an interpreting the data deluge had to be designed from the start.
equivalent time to spend with the robot. At the fair, they Our study has extended Savoia’s questions from commercial
engaged for 6.2 – 23.1 minutes (M = 14.8 mins, SD = 3.7) and product development to a framework for designing a research
most played through all four available games (M = 3.3 games, tool. The process proved useful in helping us to know what to
SD = 0.8). The children were also tolerant of the robot’s look for and identifying future needs for functionality,
occasional malfunctions. We conclude that children are demonstrating the value in extending the pretotyping concept
sufficiently interested even in the robot’s pretotype form to beyond its current setting in the commercial world.
support its research functions.
Q2. Would they buy it if we build it? Savoia’s second question REFERENCES
extends to the carers and teachers who are the gatekeepers to [1] A. Savoia. (2011, 28Nov15). Pretotype It, 2nd Ed. www.pretotyping.org
the children’s time and attention, and researchers who deploy /uploads/1/4/0/9/14099067/pretotype_it_2nd_pretotype_edition-2.pdf
the robot as a context for interaction. Key considerations in the [2] W. Buxton, Sketching user experiences: Getting the design right and the
studies concern ethics, data collection, and storage. Both right design, Morgan Kaufmann, 2007.
contexts involved voluntary visits with the robot, so it was not [3] S. Houde and C. Hill, What do prototypes prototype? Handbook of
Human Computer Interaction, 2:367-381, 1997.
surprising that all parents were positive about the research
[4] N. Dahlbak, A. Johnsson and L. Ahrenberg, Wizard of Oz studies: Why
aspects of the study. However, we were surprised by the and how? Proc Intelligent User Interfaces, ACM, 193-200, 1993.
enthusiasm shown by parents, despite the robot’s limited
[5] D. Gouaillier, +8 authors, "Mechatronic design of NAO humanoid," in
functionality and form. Testing in schools is planned to Robotics and Automation. ICRA'09. IEEE Int Conf, 2009, pp. 769-774.
observe a wider range of responses from parents and teachers. [6] M. Malmir, D. Forster, K. Youngstrom, L. Morrison, and J. Movellan,
"Home Alone: Social Robots for Digital Ethnography of Toddler
Q3. Will they use it as expected? Complementary questions are Behavior," in Proc IEEE Int Conf on Computer Vision Workshops 2013.
whether children will interact with the robot as expected and
[7] N. Weibel, S. Rick, C. Emmenegger, S. Ashfaq, A. Calvitti, and Z.
whether researchers will use the ethnography functions of the Agha, "LAB-IN-A-BOX: semi-automatic tracking of activity in the
robot as expected. The testing showed some unexpected medical office," Personal & Ubiquitous Computing, 19:317-334, 2015.
behaviors between the children and the robot with respect to [8] P. Worthy et al. "Children's Expectations and Strategies in Interacting
the tablet, voice, eye contact and physical touch. We had with a Wizard of Oz Robot" Proc 27th OzCHI 2015.
anticipated interest in the hands and had implanted touch [9] OPAL project website www.itee.uq.edu.au/cis/opal
sensors. However, children in the lab rarely touched the robot [10] K. Hensby et al, “Hand in Hand: Tools and techniques for understanding
except for the tablet and the touch sensors were not triggered. children’s touch with a social robot”, ACM/IEEE HRI 2016, in press.
At the fair, participants touched the robot on the hands, arms [11] K. Rogers et al, “Discovering Patterns of Touch: A Case Study for
and torso, and some used the frame for support (these touches Visualization-Driven Analysis in Human-Robot Interaction”,
ACM/IEEE Int Conf on Human-Robot Interaction 2016, in press.
were subsequently quantified through manual hand coding).
The cameras and microphones recorded the interactions [12] M. Evans, L. Kerlin, and C. Jay, "I Woke Up as a Newspaper:
Designing-in Interaction Analytics," in Proc 33rd ACM Conf Extended
between participants and robot in a relatively constrained field Abstracts on Human Factors in Computing Systems, 2015, pp. 477-488.

532

You might also like