Redesign of Leg Assembly and Implementation of Reinforcement Learning for a Multi-Purpose Rehabilitation Robotic Device (RoboREHAB)

• DOI: 10.3390/app14020516
• Publication Date: 2024-01-08T11:12:58Z
• Authors: Jacob Anthony; Chung-Hyun Goh; Alireza Yazdanshenas; Yong Tai Wang
• Keywords: Motion Capture; Motion analysis; Continuous passive motion

Patients who are suffering from neuromuscular disorders or injuries that impair motor control need to undergo rehabilitation regain mobility. Gait training is commonly prescribed patients muscle memory. Automated-walking devices were created aid this process; while these establish accurate ankle-path trajectories, the knee and hip movements inaccurate. In work, a redesign of leg assembly in multi-purpose robotic device (RoboREHAB) was explored improve hip- knee-movement accuracy by adding an extra link rollers assembly. Motion analysis employed test feasibility, reinforcement learning utilized train new walk, joint motions achieved with compared those motion-capture (mocap) data. As key result, motion showed improvement knee- hip-path trajectories due added roller/joint segment. The redesigned assembly, under reinforcement-learning policy, 5% deviation maximum 51.177 mm but maintained similar profile mocap trajectory This over original two-segment design, which 72.084 mm. These results hip-joint more closely reflect motion-analysis results, validating opening it up further experimentation technical improvement.