IEEE Robotics & Automation Magazine - March 2023 - 31

clinical studies, and informed consent was obtained from all
participants. The codesign principle has been a guiding line
through the project, and the comments and suggestions of the
end users have been extremely helpful to improve their experience
(e.g., by adjusting the height of a sensor, providing
additional feedback, and delivering smoother acceleration
profiles). The experimental protocols defined by the clinicians
and roboticists have been adapted to fit users' experience
(including novices, expert users, and people with
disabilities who were not allowed to operate a power wheelchair),
and scenarios of growing complexity have been proposed
to the various participants during different sessions.
The clinical trials turned out to be of paramount importance
for mechanical/electronic testing and medical validation.
DRIVING ASSISTANCE
The driving assistance solutions developed in the ADAPT project
have been tested during two clinical trials: SWADAPT1
(NCT04072536) and SWADAPT2 (NCT04259151); see
Figure 9(a) and (c). Subjects with neurological disorders participated
in these clinical studies. The main objective was to
assess their driving performance with and without assistance.
To this end, we measured the number of collisions and total
time to completion in three standardized circuits of increasing
difficulty. The SWADAPT1 clinical trial involved 25
users with expert wheelchair driving skills. The results of this
trial indicate that the proposed assistance solutions are accurate,
risk averse, and safe, with a high degree of acceptability.
Moreover, even if the participants were already expert drivers,
the study has shown that the use of the assistance module
statistically significantly reduced the number of collisions
during complex maneuvers [6]. The protocol followed during
the SWADAPT2 clinical trial was similar to that of
SWADAPT1, but 28 users with driving difficulties took part
in it. The results show a significant reduction in the number
of collisions. Notably, the more challenging the obstacle
courses are, the more useful the assistance is perceived to be.
The benefits of assistance in terms of usage and self-confidence
have been clearly demonstrated in SWADAPT2.
SpheriCol (see the " Advanced Sensing: Omnidirectional
Vision " section) has been successfully tested with patients with
cognitive disorders [19] and 17 able-bodied participants [21].
The circuit in Figure 9(b) was built in a large indoor environment
(a gymnasium) and equipped with an overhead Qualisys
motion capture system (eight Miqus cameras) to track the
wheelchair during its displacement and obtain precise ground
truth measurements for evaluation purposes. As depicted in
Figure 10, which reports a statistical analysis of the answers to
the questionnaire handed out to the 17 volunteers, SpheriCol
received neutral to positive satisfaction and encouraging
usability results from the majority of the participants. In particular,
Figure 10(d) reports the percentage of the time SpheriCol
was used by the participants in the test circuit. In addition,
even though the sample size remains relatively small, 44% of
the users engaged in our study stated that the video stream
was one of the major strengths of the driving assistant, 16%
appreciated the distance information provided by the colored
markers, and 20% deemed the system helpful for reversing the
wheelchair and for risk management (collision avoidance).
HAPTIC FEEDBACK
The haptic feedback has been evaluated with able-bodied
participants, and the clinical trials with patients were still in
progress at the time of writing. The joystick has been tested
to provide a proof of concept, and the results of this study
have been recently presented in [24]. On the other hand, the
wearable haptic armband has been assessed with healthy
(a)
(b)
(c)
FIGURE 9. The test circuits considered during the clinical trials at
(a) and (b) INSA Rennes, and at (c) Pôle Saint-Hélier.
MARCH 2023 IEEE ROBOTICS & AUTOMATION MAGAZINE
31
IEEE Robotics & Automation Magazine - March 2023

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