IEEE Robotics & Automation Magazine - March 2023 - 34

finding an optimal policy using a naive implementation based on
Bellman's principle of optimality remains a time-consuming
task. Therefore, if we are to build more granular driver models
that would be able to offer an even better fit between driving
assistance and the user's habits, this process should be accelerated.
To this end, in future versions of the shared control algorithm,
we plan to adopt a policy iteration approach.
Finally, as far as the wheelchair simulator is concerned,
we are currently considering the possibility of improving the
user experience by explicitly taking motion cues into account
(i.e., the perceptual mechanisms by which humans sense
the motion of their own body with respect to the surrounding
environment).
FUNCTIONAL CHALLENGES
While the technologies developed in the ADAPT project
have been very successful in matching the needs of the real
users identified by our clinical partners, a number of challenging
functional requirements are still missing. For example,
the assistance provided by a smart wheelchair should
always be socially acceptable, and in future iterations of our
algorithms, we are going to include an additional layer that
accommodates the social dimension (proxemics).
Moreover, in real-world scenarios, user expectations and
capabilities (e.g., the level of effort or attention) are not fixed
but subtly vary over time. To address this issue, we are currently
working on new methods that dynamically adapt the
level of assistance to the instantaneous needs of the user. For
that purpose, we intend to take advantage of an eye tracker
and body sensors to monitor the physiological and biochemical
profile of the driver in the short and long terms (in fact,
biomarkers in saliva and sweat are known to be indicative of
performance and stress).
RECOMMENDATIONS
As the five-year term of the ADAPT project comes to an end,
it is certainly worthwhile here to sum up some of the key
findings and lessons learned, based on our own experience of
the terrain. These guidelines are intended for researchers in
rehabilitation and assistive robotics and for health-care professionals.
They are as follows:
■ The development of a new smart wheelchair requires the
concerted effort of three actors throughout the process
( " codesign principle " ): medical specialists, robotic
researchers, and end users. A mere transfer of consolidated
robotic technologies is doomed to failure.
■ Simplicity, modularity, and ergonomics are fundamental
design principles for smart wheelchairs, and they cannot
be sacrificed in the development stage.
■ Haptic interfaces are emerging assistive devices for power
wheelchairs. They are minimally invasive and intuitive to
use, but they still have not found their way into mainstream
clinical practice today. Likewise, omnidirectional
vision has not met with widespread acceptance.
■ Training programs for health-care professionals to
learn new assistive technologies ( " train-the-trainer "
34 IEEE ROBOTICS & AUTOMATION MAGAZINE MARCH 2023
sessions) are crucial to accelerate deployment toward
full-scale adoption.
■ The journey to the market is long and arduous (especially
in the time of COVID-19). For instance, the time elapsed
between the submission of the experimental protocol and
approval by the local ethics committee can exceed the
length of the product development phase.
CONCLUSION
This article provided a general overview of the innovative
assistive robotic technologies developed in the ADAPT project.
The exposition focused on the design, implementation,
and experimental validation, via large-scale clinical trials, of
two complementary smart wheelchairs and a wheelchair
driving simulator based on virtual reality. This research, carried
out by an international team of roboticists and medical
experts, is rooted in two basic principles, codesign and modularity,
and it has the potential to transform the everyday life
of millions of wheelchair users worldwide.
ACKNOWLEDGMENTS
This work was carried out as part of the Interreg VA France
(Channel) England (FCE) ADAPT (Assistive Devices for
Empowering Disabled People Through Robotic Technologies)
project. The Interreg FCE program is a European Territorial
Cooperation program that aims to fund high-quality
cooperation projects in the Channel border region between
France and England. The program is funded by the European
Regional Development Fund (ERDF). The authors would like
to thank Éric Bazin (Institut d'Electronique et des Technologies
du Numérique, INSA Rennes) for his help during the
development of the hardware platforms and the Inria associate
team ISI4NAVE for supporting several exchanges among
the French and English partners.
AUTHORS
Fabio Morbidi, Modeling, Information, and Systems laboratory,
University of Picardie Jules Verne, Amiens, 80039,
France. Email: fabio.morbidi@u-picardie.fr.
Louise Devigne, Rainbow/Hybrid teams, Institut de
Recherche en Informatique et Systèmes Aléatoires/Inria Rennes
and Institut National des Sciences Appliquées de Rennes,
Rennes, 35700, France. Email: louise.devigne@irisa.fr.
Catalin Stefan Teodorescu, Aspire Create, University
College London, London, WC1E 6BT, U.K., and Royal
National Orthopedic Hospital, London, HA7 4LP, U.K.
Email: s.teodorescu@ucl.ac.uk.
Bastien Fraudet, Pôle Saint-Hélier, Rennes, 35000,
France. Email: bastien.fraudet@pole-sthelier.com.
Émilie Leblong, Pôle Saint-Hélier, Rennes, 35000,
France. Email: emilie.leblong@pole-sthelier.com.
Tom Carlson, Aspire Create, University College London,
London, WC1E 6BT, U.K., and Royal National Orthopedic
Hospital, London, HA7 4LP, U.K. Email: t.carlson@ucl.ac.uk.
Marie Babel, Rainbow/Hybrid teams, Institut de
Recherche en Informatique et Systèmes Aléatoires/Inria
IEEE Robotics & Automation Magazine - March 2023

Table of Contents for the Digital Edition of IEEE Robotics & Automation Magazine - March 2023
