부산대학교 도서관

Soft exoskeleton robots have exhibited promising potential in walking assistance with comfortable wearing experience. However, traditional cable-driven exosuits cannot provide adequate driving force to motivate the entire human leg, especially for hemiplegic patients with little movement capability. Also, the human-exosuit coupling dynamics is difficult to be modeled due to the suit-like structure and the varying human performance, and accordingly, accurate control and efficient assistance cannot be guaranteed. In this article, twisted string actuators (TSAs) are developed and equipped with the exosuit to provide powerful actuation and variable assistance intensity. Besides, the human motion intention is estimated based on skin surface electromyography (EMG) signals. A mirror adaptive impedance control is proposed, where the control torques and stiffnesses of the TSAs are regulated based on the performance of the impaired limb and the motion reference of the healthy limb. A linear quadratic regulator (LQR) is formulated to minimize the movement trajectory tracking errors and the human physical effort. An integral reinforcement learning algorithm is adopted to solve the given LQR problem to optimize the impedance parameters with little information of the human and robot models. The proposed robotic system is validated through experiments to perform its effectiveness and superiority.