Validation of Fine Manipulation Using NMPC for Rotation Floating Space Robots with HILS Setup

On-orbit servicing using autonomous robotic manipulators onboard satellites is a promising method for future space sustainability missions like debris removal, refueling, assembly, and maintenance. A newer variety of space robots is the Rotation floating space robot (RFSR), in which the orientation and position of a robotic arm mounted on a floating satellite are actively controlled. At the same time, the satellite base orientation is also being simultaneously adjusted. However, non-linearities and complexities associated with controlling such coupled space-based systems present difficulties in their feasible implementation, requiring accurate modeling and advanced control algorithms. It is challenging to model such systems with coupled degrees of freedom (DOF) and a high degree of non-linearity. The paper presents an approach to model and control Rotation Floating Space robots to design an optimal path that the end-effector can follow while being controlled to capture the target. A Non-linear Model Predictive Controller (NMPC) is proposed for the position and orientation control of a UR5 manipulator mounted on the base of the satellite. Also, experimental verification of such systems is difficult due to difficulties replicating a micro-gravity environment on the ground, providing 6DOF motion to robots. This paper addresses multiple issues. One is to model RFSR in Pybullet and use NMPC-driven layered control architecture for simultaneous 9DOF control. The simulation shows the complete workflow from deploying the robotic arm until the target is successfully captured or docked on. After that, we use a hardware-in-the-Ioop (HILs) hybrid approach to validate the simulations performed in PyBullet. While one robot is employed to simulate the target body's six degrees of freedom, another robot mimics the movements of the chaser arm. The NMPC formed the outer loop controller, generating reference commands for the inner loop PyBullet controller. The NMPC took the gripper to the desired pose while a simultaneous PD controller stabilized the orientation of the satellite base and maintained it at the desired values. The results from simulation and hardware experiments show that the proposed controllers were efficient, and the NMPC could find an optimal path to the desired gripper's pose.