Abstract
A systematic approach for the supervision-based safe motion control of mobility assistive robots based on invariance control is presented. It allows the formulation of safety features in the form of a constraint admissible state space region that can be kept invariant by proper switching between a nominal and a corrective controller whenever a predefined safety constraint is about to be in violation. Boundaries on the human-robot distance are considered critical safety features for human forward fall and human-robot collision avoidance. The nominal and corrective controller as well as the switching policy are derived. The approach is validated in simulations and experiments and shows high potential for building a systematic safety framework for mobility assistive robots.