AUTHOR=Zhao Sikai , Zheng Tianjiao , Sui Dongbao , Zhao Jie , Zhu Yanhe 

TITLE=Reinforcement learning based variable damping control of wearable robotic limbs for maintaining astronaut pose during extravehicular activity

JOURNAL=Frontiers in Neurorobotics

VOLUME=Volume 17 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/fnbot.2023.1093718

DOI=10.3389/fnbot.2023.1093718

ISSN=1662-5218

ABSTRACT=As astronauts performing on-orbit service of extravehicular activity (EVA) without the help of space station robotic arms, it will be rather difficult and labor-consuming to maintain the appropriate posture in case of impact. In order to solve this problem, a wearable robotic limbs system for astronaut assistance is developed and its variable damping control method for posture maintenance is proposed. The requirements of the astronaut’s impact resisting ability during EVA are analyzed, including the capabilities of deviation resistance, oscillation resistance, fast return and accurate return. To meet these needs, the system of the astronaut with robotic limbs is modelled and simplified. In combination with this simplified model and reinforcement learning algorithm, a variable-damping controller for the end of the robotic limb is obtained, which can regulate the dynamic performance of the robot end to resist oscillation after impact. A weightless simulation environment for the astronaut with robotic limbs is constructed. The simulation results demonstrate that the proposed method can meet the proposed pose maintaining requirements. In comparison to the fixed damping control method, the variable damping controller has better performance in preventing excessive deviation from the original position and fast return to the starting point. The maximum deviation displacement is reduced by 39.3% and the recovery time is cut by 17.7%. Besides, it can improve the accuracy of the return to the original position and prevent the reciprocating oscillation.