基于强化学习的软体机构抓捕策略研究

大型空间结构建造与维护、失效卫星检测与维修、轨道碎片清除等已成为航天技术发展亟待解决的现实问题。针对传统空间捕获机构质量惯量大、末端抓取精度要求高、抓捕对象适用范围窄不足等,创新性地提出基于IPMC(ion-exchange polymer metal composite)功能复合材料的多自由度仿生软体新型抓捕机构,同时基于强化学习算法提出多模态信息融合的抓捕操作强化学习策略,从而提升抓捕机构空间捕获的成功率,为空间抓捕技术的智能化发展提供新思路。

Research on the Grip Strategy of Rheid Institutions Based on Q-Learning

The large space structure construction and maintenance, failure satellite inspection and the orbital debris removal have become the practical problems which should be solved. In view of the great quality inertia of capturing mechanism, high end-grip accuracy requirement and narrow scope for grip the object in traditional space, multiple freedom degrees bionic rheid grip institution based on IPMC(ion-exchange polymer metal composite) functional composite materials grips is put forward innovatively, at the same time, based on the Q-Learning algorithm, the grip operation reinforcement learning strategy of multi-mode information fusion is proposed, so as to improve the success rate of space grip. This study provides a new idea for the intelligent development of spatial grip technology.