A learning system for motion planning of free-float dual-arm space manipulator towards non-cooperative object

In recent years, non-cooperative objects, such as failed satellites and space junk, can be detected in space. These objects are usually manipulated or collected by free-floating dual-arm manipulators. Reinforcement learning methods show more promise trajectory planning for manipulators difficulties modeling manual parameter tuning have been recently surmounted. Although previous studies demonstrated their effectiveness, they cannot applied to track dynamic targets with unknown rotation (non-cooperative objects). this paper, we proposed a system manipulator motion against objects. Specifically, our method consists of two modules. Module I realizes the multi-target end-effectors within large target Next, II takes point clouds object input estimate motional properties, then location points on predicted. Target rotating successfully tracked through combination II. Furthermore, experiments also demonstrate scalability generalization system.