MECHANICAL SIMULATION AND FULL ORDER SLIDING MODE COLLISION AVOIDANCE COMPLIANT CONTROL BASED ON NEURAL NETWORK OF DUAL-ARM SPACE ROBOT WITH COMPLIANT MECHANISM CAPTURING SATELLITE

The problem of collision avoidance compliance control for dual-arm space robot to protect joint due to impact in the process of capturing satellite is discussed. For this reason, a rotatory series elastic actuator (RSEA), a compliant mechanism, is designed between the joint motor and the manipulator. It has two functions: firstly, the impact energy of satellite to robot joints can be absorbed by RSEA's built-in spring through stretching or compressing in the capture operation; secondly, the impact torque of the joints can be limited in the safe range by reasonably designing a matching collision avoidance compliance control strategy. First of all, the dual-arm space robot with compliant mechanism open-loop subsystem dynamics model and the target satellite subsystem dynamics model are established before capture operation by the second Lagrange equation. Then, based on the momentum conservation and geometric constraints of the position and velocity of the closed-chain system, the closed-chain hybrid system of the space robot and the captured satellite is obtained after the capture operation. Finally, for calm control the hybrid system, based on RBF neural network, a full-order terminal sliding mode collision avoidance compliance control scheme is proposed. The proposed scheme not only can effectively absorb and buffer the impact energy in the capture operation, but can turn on or off the space robot's joint motor timely when the impact energy is too large, so as to avoid overload and damage of the joint actuator. In addition, the joint torques are allocated by the minimum weight norm theory to ensure the coordinated operation between manipulators. The global stability of the system is proved by the Lyapunov theory. At last, the effectiveness of the collision avoidance compliance control strategy is verified by computer simulation.