A control theoretic approach to robot-assisted locomotor therapy

This paper proposes a control theoretic strategy for human walking gait assistance based on underactuated potential energy shaping. We design a simple control law that lessens the perceived weight of the patient's center of mass through a robotic ankle-foot orthosis (AFO) with one actuated degree-of-freedom. We then adopt a passive “compass-gait” bipedal walker as an implicit model of human locomotor behavior, which we simulate to draw beneficial implications for rehabilitation such as energy regulation, improved stability, and progressive training by Lyapunov funneling. Given current challenges in developing effective robot-assisted locomotor therapies, this paper offers a novel systematic approach to control strategy design for gait training and at-home assistance.