Constrained Trajectory Planning and Control for Space Manipulator Capture Tasks Using an Improved Bilevel MOPSO Algorithm

Abstract

This article tackles the joint constrained trajectory planning (CTP) issue for capture tasks of space manipulator subject to multiple joint state constraints and multiple optimization objectives. By parameterizing the joint trajectory using polynomial spline curves, the original CTP problem is reformulated as a bilevel constrained multiobjective optimization (BLCMO) problem. This formulation explicitly incorporates kinematic constraints on joint angles (including the initial configuration, terminal configuration, and intermediate waypoints), joint angular velocities, and joint angular accelerations, while simultaneously minimizing the total motion time and a jerk-based trajectory smoothness index. To solve the resulting BLCMO problem, an improved bilevel constrained multiobjective particle swarm optimization (CMPSO) algorithm is constructed, which combines a gradient-assisted CMPSO scheme at the upper level optimization problem with a quadratic programming (QP) algorithm at the lower level optimization problem. In this algorithmic framework, the upper level CMPSO searches over the motion time sequence, whereas the lower level QP obtains the spline coefficient vector, thereby jointly and significantly reducing the search dimensionality and computational complexity of the original CTP problem. A low-level model-based dynamic control strategy employing a fully actuated system approach is used to stabilize the spacecraft platform and track the planned joint motion trajectory. Finally, comparative numerical simulations with a seven-degree-of-freedom space manipulator and hardware experiments with the Franka manipulator are performed to demonstrate the performance of the proposed scheme.