Position/Force Control for a Dual-Arm Space Manipulator Gripping and Transporting an Unknown Target Without Grapple Fixtures

This article proposes a position and force control method for a dual-arm space manipulator to grip and transport an unknown target without grapple fixtures. The hierarchical sliding mode controller drives the end-effectors to desired positions and attitudes in the workspace and limits the joints' displacements in certain ranges simultaneously, which has the advantage of avoiding self-collision and singularities. The force controller is designed based on reinforcement learning instead of modeling the dynamic contact process or the usage of the force senors, which regulates the contact force by training a proximal policy optimization agent to grip the unknown target without grapple fixtures during transportation. A physical simulation system is built for the dual-arm space manipulator operation system. Numerical simulations are carried out to verify the applicability and effectiveness of the proposed control method.