On-orbit assembly control based on model predictive control and load online identification

Abstract

Space manipulator are essential for on-orbit assembly (OOA) tasks. This paper proposes an advanced control framework to address key challenges in Cartesian trajectory tracking and compliant assembly. First, a rigid-body dynamic model of the manipulator is established using CAD model and joint friction identification. A Linear Extended State Observer (LESO) is designed to mitigate system errors by modeling both parametric uncertainties and external disturbances as an extended state. Additionally, the method for selecting parameters of the LESO, along with its convergence proof, is presented. Second, a model predictive control (MPC) algorithm is employed to optimize the Cartesian trajectory of the manipulator in real time. By linearizing the Cartesian space trajectory, the MPC problem is formulated as a QP problem. Solving this QP problem in real-time allows for the optimization of the Cartesian space trajectory, thereby improving the trajectory tracking accuracy in Cartesian space. Finally, online identification of load parameters, sensor zero drift, and temperature drift enables effective compliant assembly control. Extensive experiments validate the proposed methods: (1) The Cartesian trajectory tracking error is maintained below 0.1 mm. (2) After sensor compensation for identified load parameters, zero drift, and temperature drift, the maximum force residual is 0.42 N, and the maximum torque residual is 0.1 N m.