Machine learning and unlearning to autonomously switch between the functions of a myoelectric arm

Abstract

Powered prosthetic arms with numerous controllable degrees of freedom (DOFs) can be challenging to operate. A common control method for powered prosthetic arms, and other human-machine interfaces, involves switching through a static list of DOFs. However, switching between controllable functions often entails significant time and cognitive effort on the part of the user when performing tasks. One way to decrease the number of switching interactions required of a user is to shift greater autonomy to the prosthetic device, thereby sharing the burden of control between the human and the machine. Our previous work with adaptive switching showed that it is possible to reduce the number of user-initiated switches in a given task by continually optimizing and changing the order in which DOFs are presented to the user during switching. In this paper, we combine adaptive switching with a new machine learning control method, termed autonomous switching, to further decrease the number of manual switching interactions required of a user. Autonomous switching uses predictions, learned in real time through the use of general value functions, to switch automatically between DOFs for the user. We collected results from a subject performing a simple manipulation task with a myoelectric robot arm. As a first contribution of this paper, we describe our autonomous switching approach and demonstrate that it is able to both learn and subsequently unlearn to switch autonomously during ongoing use, a key requirement for maintaining human-centered shared control. As a second contribution, we show that autonomous switching decreases the time spent switching and number of user-initiated switches compared to conventional control. As a final contribution, we show that the addition of feedback to the user can significantly improve the performance of autonomous switching. This work promises to help improve other domains involving human-machine interaction - in particular, assistive or rehabilitative devices that require switching between different modes of operation such as exoskeletons and powered orthotics.