The effect of robotic wheelchair control paradigm and interface on user performance, effort and preference: An experimental assessment

The exact manner in which control is shared between a human and an autonomous system is a crucial factor for assistive robots that provide physical support to people with severe motor impairments. There has however been little comparative study between different control-sharing paradigms within the field of assistive robotics. We present a control architecture for a robotic "smart" wheelchair that allows for the seamless interchange and evaluation of any number of control-sharing paradigms, and so facilitates comparative study between them. We present an implementation of four control sharing paradigms, and results from a study that compares all four to each other and teleoperation, and moreover using multiple control interfaces and across multiple sessions. Experimental results suggest that (i) task performance metrics differ with each control interface, (ii) performance increases with increasing autonomy assistance however it is not statistically significantly different between higher levels of autonomy, (iii) metrics related to user effort show a decrease with increasing autonomy, which is more emphasized with more limited control interface and (iv) how much the autonomy is utilized differs greatly between control paradigms, but not control interfaces. Moreover, (v) for almost all performance metrics, there is a consistent performance increase in Session 2 compared to Session 1, and for both control interfaces. Lastly, subjective questionnaires (control paradigm preference and perceived utility) reveal both similarities and differences between SCI and uninjured subjects. No single control paradigm is the clear winner in performance or preference, suggesting that it will be important to offer end-users multiple control options to accommodate their individual needs and preferences. (C) 2017 Elsevier B.V. All rights reserved.