Shaping Individualized Impedance Landscapes for Gait Training via Reinforcement Learning

Assist-as-needed (AAN) control aims at promoting therapeutic outcomes in robot-assisted rehabilitation by encouraging patients' active participation. Impedance is used most AAN controllers to create a compliant force field around target motion ensure tracking accuracy while allowing moderate kinematic errors. However, since the parameters governing shape of are often tuned manually or adapted online based on simplistic assumptions about subjects' learning abilities, effectiveness conventional may be limited. In this work, we propose novel adaptive controller that capable autonomously reshaping phase-dependent manner according each individual's motor abilities and task requirements. The proposed consists modified Policy Improvement with Path Integral algorithm, model-free, sampling-based reinforcement method learns subject-specific impedance landscape real-time, hierarchical policy parameter evaluation structure embeds paradigm specifying performance-driven goals. adaptability strategy responses its ability promote short-term adaptations experimentally validated through treadmill training sessions able-bodied subjects who learned altered gait patterns assistance powered ankle-foot orthosis.