Toward Seamless Transitions Between Shared Control and Supervised Autonomy in Robotic Assistance

Assistive robots aim to help humans with impairments execute motor tasks in everyday household environments. Controlling the end-effector of such directly, for instance a joystick, is often cumbersome. Shared control methods, like Control Templates (SCTs) [1], have therefore been proposed provide support robotic control. Moreover, depending on factors as workload, system trust or engagement, users may freely adjust level autonomy, by letting robot complete task itself. In this letter, we present concept adjustable autonomy context assistance. We extend SCT approach an automatic module that allows user switch between and Supervised Autonomy at any time during execution. As both modes use same action representation, transitions are seamless. show capabilities set daily living our wheelchair-mounted EDAN humanoid Rollin' Justin. highlight how execution benefits from features, task-related constraints whole-body