Computing Conference 2017

18-20 July 2017 | London, UK

Use of Continuum Robots for Remote Inspection

Operations
Ian D. Walker, Clemson University
Department of Electrical and Computer Engineering
320 Fluor Daniel Building
Clemson, South Carolina 29634 USA
iwalker@clemson.edu

Abstract—We discuss the potential of long, thin “continuum”
robots to enable new and more effective approaches to remote
inspection. Resembling robotic cables, this new class of robots
can enter and explore congested and potentially unstable
environments, sending back information from sensors at their
tips. This capability is of particular value in search operations in
disaster relief situations. We illustrate the potential using a thin
continuum robot in representative tasks.
Keywords—robotics; disaster relief; continuum robots
I. PURPOSE
In many human or robotic incursions into unknown or
partially known environments, and notably in disaster relief
operations, a priori remote sensing and inspection is invaluable
and often key to success. In urban disaster relief situations,
lives can depend on whether the existence and location of
buried victims and/or dangerous entities (gas leaks, explosive
materials, etc.) can be established among damaged
infrastructure. Further, inspection needs to be made without
further collapsing structures within the environment [8].
There is therefore a need for alternative sensor placement
technologies which can maneuver through tight space in
cluttered, complex, a priori unknown (or partially known)
environments. In the event of contact between the deployed
technology and its surroundings (either planned or
inadvertent), the machine interface needs to be compliant, to
prevent the generation of high contact forces which could
destabilize the environment.
Conventional robot technologies are based around rigid
elements (links, wheels, tracks,…), which inherently present a
relatively high stiffness mechanical interface to the
environment. While this is highly advantageous in traditional
robot application arenas (factories, hard floors/road surfaces,
etc.), enabling high precision and repeatability operations in
structured or semi-structured environments, it is less inherently
suited to compliant and adaptive operation in unstructured and
potentially unstable environments.
Our work is focused on the development and evaluation of
an alternative approach to remote robotic sensor placement in
unstructured environments. The approach is focused on the use
of a low profile, highly maneuverable, compliant emerging
robot technology: continuum robots. The key goal of the work
is to evaluate the potential of such robot structures to provide
an alternative and effective means to deploy sensors in a priori
unknown cluttered environments.
II. BACKGROUND/SIGNIFICANCE
Continuum robots [9] are a novel and rapidly emerging
class of robot [11],[14] with continuously bendable backbones.
Sometimes inspired by biological structures such as elephant
trunks [5], octopus arms [6],[15] and vines [16], continuum
robots are inherently more compliant and adaptable than
conventional robot structures based on rigid links. This
compliance allows them to gently maneuver amongst and
through obstacles, while avoiding the generation of large
contact forces.
Herein, we discuss long (length measured in meters), and
thin (length two orders of magnitude or more than diameter),
cable-like continuum robots aimed at remote inspection
operations [10],[13]. The large length (multiple meters) to
diameter (less than two centimeters) ratio of these robots, along
with their ability to bend continuously along their length,
allows deep penetration into congested spaces, such as debris
from collapsed buildings.
Continuum robots have been proposed previously for
inspection of nuclear [1],[2],[4] and space [7],[16]
environments. Thin (but not long, in the sense of multiple
meter length) continuum robots have found a niche in medical
applications [3]. However, apart from some early efforts [12],
in which the proposed robot was not long and thin in the sense
discussed herein, continuum robots have not been utilized in
disaster relief operations.
We next illustrate and detail the potential for long thin
continuum robots for remote inspection using as an example a
prototype long thin continuum robot, equipped with a small
camera at its tip. We discuss the capabilities and operation of
the robot, and show examples of its deployment within, and
remote investigation of, congested spaces.
III. METHOD
A. Continuum Robot Design
The prototype long, thin (length ~2m, diameter ~0.07m)
continuum robot used in the examples herein is shown in Fig.
1. The flexible backbone consists of three concentrically fitted
carbon fiber tubes, spring-loaded to allow local
extension/contraction as well as two-dimensional bending of

1382 | P a g e
978-1-5090-5443-5/17/$31.00 ©2017 IEEE

each section. The robot is remotely actuated by nine electric
servomotors, which pull tendons. Sets of three tendons, spaced
at 120 degrees apart as viewed from the backbone cross-
section, are terminated at each of the three sections. The
actuators are mounted in a compact assembly (not shown) at
the base of the robot. A small USB camera was mounted at the
tip of the robot.
Fig. 1. Thin tendril continuum robot (length: over two meters, maximum
diameter less than one centimeter)
The three independently controlled sections of the robot,
each with three degrees of freedom, result in a nine
controllable degree of freedom system. The bending stiffness
of the tubes forming the backbone, and the spring stiffnesses
determining the compressibility of the structure were selected
to produce a system whose unloaded shape is easily
controllable, but which is highly compliant to externally
applied loading. Full design and operational details of the robot
are given in [16].
B. Operation of the Robot
The robot was deployed in two types of test environments,
detailed in the following section. The first task (section IV.A)
focused on finding a known object (a cell phone) in an a priori
unknown and collapsible indoor environment. The second task
focused on exploring and categorizing elements of an a priori
unexplored outdoor environment. In each case the ability of the
robot to manoeuver through the obstacle field while contacting
but not collapsing the environment was recorded.
Coarse base mobility for the robot was enabled manually
(i.e. a human carried the robot to the boundary of each
environment, and selected the initial direction in which to
insert the tendril). The tendons were then controlled to for fine
navigation of the tendril through the obstacles in each
environment, guided by the operator using the video from the
tip camera. The camera image was the only feedback provided
to the human operator.
IV. RESULTS
In the example deployments below, neither environment
was modified for the reported experiments beyond the hiding
978-1-5090-5443-5/17/$31.00 ©2017 IEEE
Computing Conference 2017
18-20 July 2017 | London, UK
of a specific object to be searched for within the first
environment.
A. Searching in an priori Unknown Cluttered Space
We deployed the robot in Fig. 1 in a series of congested
interior spaces. Fig. 2 shows a tip camera view during the
deployment of the robot in a specific interior space, the interior
of a recycling bin. In this experiment, the robot entered through
a hole at the top, and was remotely operated to locate a hidden
object (a cell phone) among the objects within the bin.
Fig. 2. Robot deployed among dense unstructured objects. Cell phone target
is found “alive” (i.e. on) behind cans, paper, etc
The contents of the bin mostly comprised cans and bottles,
mixed in with paper and cardboard pieces (typically folded
and/or crumpled). This environment was particularly
challenging due to the fact that the generation of high contact
forces between the robot and the paper/cardboard tended to
collapse those environmental elements, significantly changing
the geometry of the environment, and blocking paths to the
target Thus this environment was a good opportunity to
evaluate the ability of the inherent compliance within the
tendril robot to gently explore potentially unstable
environments.
A representative sequence of images from the tip camera
during the experiment are shown in Fig. 3. The robot was able
to negotiate the clutter without collapsing environmental
obstacles, locate the cell phone, and confirm that it was “alive”
(i.e. on). In this context, the mechanical potential and
capabilities of the tendril robot were validated. However, the
tip camera view of the environment proved confusing at times
for the operator, who tried multiple times to direct the tendril
through the same impassible regions of the bin, before
ultimately finding a feasible path. This observation motivates
the development of sensor-based mapping and visualization
techniques to augment these types of searches. Note the
analogy in this experiment to searching in cluttered
environments.
1383 | P a g e
 Computing Conference 2017
18-20 July 2017 | London, UK

Fig. 3. Tip camera sequence showing tendril search. Top: point of entry
from top of container. Middle: typical internal view from robot tip – clearly
identifiable cans, largely occluded by paperr and cardboard. Bottom: ultimate
emegence into center of container. Visible (immediate top right of red soda
can) is search goal, a cell phone. Cell phone is seen to be on, i.e. “alive”

B. Exploring Outside in Natural/Artificial Environment

The robot was also deployed in outdoor environments. In
Figs. 4 and 5, tip camera views of one deployment of the robot
through dense undergrowth to locate, identify, and enter a drain
and explore its contents are illustrated. This experiment was
performed at night, in fairly dense vegetation. Unknown to the
robot operator was the existence of a drain (with a cover grill
over the entry), obscured by the undergrowth. The planned Fig. 4. Tip camera sequence showing tendril exploration through vegetation,
goal of the experiment was for the operator to explore, locate ultimately finding and approaching drain cover. Top: drain cover becomes
and identify the drain, and then enter and explore through the initially visible through vegetation. Middle images: drain cover and its
drain cover and down the drain with the tendril. However, geometry become increasingly clear as robot bends around and through
vegetation, approaching drain. Bottom: closer approach, as tendril tip bends to
unknown to the research team as a whole was that the drain
begin entry into drain
was filled with water to a short depth below the drain cover.

1384 | P a g e
978-1-5090-5443-5/17/$31.00 ©2017 IEEE

Fig. 5. Close up of (reflection of) robot entering through drain cover. Image
(from tip camera) reflected from water in drain. Robot tip can be seen (left)
entering drain cover opening
The robot was deployed into the environment from above,
and was able to bend along the backbone to move within and
through the vegetation. On reaching the ground, the tip section
was bent and rotated to locate the drain, and then reconfigured
to gain entry through a small entry port. On entry into the
drain, the camera images successfully identified the layer of
water approximately 40 cm below the drain cover.
A representative sequence of images from the tip camera
during the experiment are shown in Fig. 5. The experiment
demonstrated the ability of the tendril to negotiate a non-trivial
combination of natural and engineered clutter, and to discover
and identify a priori unknown environmental conditions. Note
the analogy in this experiment to searching for dangerous
chemicals.
V. CONCLUSIONS
We have discussed the potential of long, thin, cable-like
continuum robots to provide a new capability in remote
inspection. Via their ability to penetrate and explore congested
environments, such robots can deliver sensor payloads to
locations beyond the reach of conventional devices or robots.
Their inherent compliance further enables them to explore
these environments “gently”, avoiding the generation of large
contact forces which might destabilize sensitive environments.
These novel capabilities match well to the needs of remote
inspection in disaster relief situations, as well as environmental
monitoring applications.
The potential of the underlying concept is illustrated using
a prototype long thin continuum robot in representative tasks.
We illustrate the results of two representative experiments with
a prototype tendril robot, within natural and human-created
indoor and outdoor environments. The robot was shown to be
able to successfully navigate and identify target objects.
978-1-5090-5443-5/17/$31.00 ©2017 IEEE
Computing Conference 2017
18-20 July 2017 | London, UK
ACKNOWLEDGMENT
The author would like to thank Phanideep Gonthina and
Ryan Scott for conducting the experiments and obtaining the
images used in this paper.
REFERENCES
[1] R.V. Bostelman, J.S. Albus, and R.E. Graham, “Robocrane and EMMA
Applied to Waste Storage Tank Remediation”, Proceedings American
Nuclear Society Seventh Topical Meeting on Robotics and Remote
Systems, Augusta, GA, pp. 708-713, 1997.
[2] R. Buckingham, “Snake Arm Robots,” Industrial Robot: An
International Journal, vol. 29, no. 3, pp. 242-245, 2002.
[3] J. Burgner-Kars, D.C. Rucker, and H. Choset, “Continuum Robots for
Medical Applications: A Survey”, IEEE Transactions on Robotics, Vol.
31, No. 6, December 2015, pp. 1261-1280.
[4] R.E. Graham and R.E. Bostelman, “Development of the EMMA
Manipulator for Hazardous Waste Remediation”, Proceedings American
Nuclear Society Seventh Topical Meeting on Robotics and Remote
Systems, Augusta, GA, pp. 646-650, 1997.
[5] M. W. Hannan and I. D. Walker, “Kinematics and the Implementation of
an elephant's trunk manipulator and other continuum style robots,”
Journal of Robotic Systems, vol. 20, no. 2, pp. 45-63, Feb. 2003.
[6] W. McMahan, M. Pritts, V. Chitrakaran, D. Dienno, M. Grissom, B.
Jones, M. Csencsits, C.D. Rahn, D. Dawson, and I.D. Walker, “Field
Trials and Testing of “OCTARM” Continuum Robots”, Proceedings
IEEE International Conference on Robotics and Automation, pp. 2336-
2341, 2006.
[7] J.S. Mehling, M.A. Diftler, M. Chu, and M. Valvo, “A Minimally
Invasive Tendril Robot for In-Space Inspection”, Proceedings
BioRobotics 2006 Conference, pp. 690-695, 2006.
[8] R.R. Murphy, S. Tadokoro, D. Nardi, A. Jacoff, P. Fiorini, H. Choset,
and A.M. Erkmen. "Search and Rescue Robotics", in Springer
Handbook of Robotics, pp. 1151-1173. Springer Berlin Heidelberg,
2008.
[9] G. Robinson and J. B. C. Davies, “Continuum Robots - A State of the
Art,” Proceedings IEEE International Conference on Robotics and
Automation, Detroit, Michigan, pp. 2849-2854, 1999.
[10] M. Tonapi, I.S. Godage, A.M. Vijaykumar, and I.D. Walker, “Spatial
Kinematic Modeling of a Long and Thin Continuum Robotic Cable”,
Proceedings IEEE International Conference on Robotics and
Automation, Seattle, WA, May 2015, pp. 3755-3761.
[11] D. Trivedi, C.D. Rahn, W.M. Kier, and I.D. Walker, “Soft Robotics:
Biological Inspiration, State of the Art, and Future Research”, Applied
Bionics and Biomechanics, 5(2), pp. 99-117, 2008.
[12] H. Tsukagoshi, A. Kitagawa, and M. Segawa, “Active Hose: An
Artificial Elephant's Nose with Maneuverability for Rescue Operation,”
Proceedings IEEE International Conference on Robotics and
Automation, Seoul, Korea, pp. 2454-2459, 2001.
[13] I.D. Walker, “Robot Strings: Long, Thin Continuum Robots”,
Proceedings IEEE Aerospace Conference, Big Sky, MT, pp. 1-12, 2013.
[14] I.D. Walker, “Continuous Backbone “Continuum” Robot Manipulators:
A Review”, ISRN Robotics, Vol. 2013, July 2013, pp. 1-19.
[15] I.D. Walker, D. Dawson, T. Flash, F. Grasso, R. Hanlon, B. Hochner,
W.M. Kier, C. Pagano, C.D. Rahn, Q. Zhang, “Continuum Robot Arms
Inspired by Cephalopods, Proceedings SPIE Conference on Unmanned
Ground Vehicle Technology VII, Orlando, FL, pp. 303-314, 2005.
[16] M.B. Wooten and I.D. Walker, “A Novel Vine-Like Robot for In-Orbit
Inspection”, Proceedings 45th International Conference on
Environmental Systems, Bellevue, WA, July 2015, pp. 1-11.
1385 | P a g e