Tips：If you see a small figure in the center of the above screen, click it and then you will see the video.

we developed a hierachical decoupling optimization algorithm to realize 6D-pose multi-target trajectory planning for the free-floating space robot.

Abstract

The free-floating space robot plays an improtant role in recycling space junk and reparing satellite . Because of the coupling between the base and robot, there is a dynamic singularity problem in trajectory planning. Many researches focus on parameterizing the trajectory and proposing optimization-based algorithms to control the space robot. However, the previous solutions will be faild when the target to be captured has some movement. So these methods need recalculte a complete trajectory involving new pose of the target. In this paper, we developed a model-free Hierarchical Decoupling Optimization (HDO) algorithm to realize 6D-pose multi-target trajectory planning for the free-floating space robot. In order to reduce the complexity of exploration, the whole system consists of two layers: the high-level policy completes the collision-free trajectory planning of the pose of end-effector; the low-level policy divides the arbitrary pose reaching task into two decoupling sub-tasks (position and orientation) within a large target space. Thanks to introducing the Hindsight Experience Replay (HER), we successfully trained two agents based on multi-goal reinforcement learning. In particular, we proposed an Event-based Asynchronous Optimization (EAO) , which facilitates our method to efficiently approximate the optimal policy. Theoretical analysis shows EAO can guarantee the learning stability and reachable global optimal point. The simulation results illustrate that the proposed algorithm achieves high environmental adaptability and anti-disturbance capacity. Furthermore, our approach generalizes to capture a target satellite, which is a contact-rich environment.