Obstacle Avoidance of Dual-Arm Space Robots via Nonlinear Model Predictive Control and Velocity Dampers

Abstract

Compared with a single-arm configuration, a dualarm space robot can share grasping loads, improve capture stability, and provide mutual backup during non-cooperative target operations. This paper addresses the coordinated pre-capture of a spinning non-cooperative target by a dual-arm free-flying space robot considering obstacle-avoidance constraints. A three-dimensional kinematic model is established for the coupled base attitude, joint variables, and dual end-effector motion. Piecewise reference trajectories guide the end-effectors toward the two grasping points on a spinning cylindrical target. To avoid the cost of the full high-dimensional nonlinear model predictive control (NMPC), a velocity-level implementation is adopted while retaining receding correction, coordinated allocation, and constraint handling. Damped least squares improve the robust performance near ill-conditioned Jacobian configurations, and the velocity damper method converts obstacle-clearance constraints into linear inequalities on end-effector normal velocity. Sequential projection then corrects the nominal joint velocity online. Numerical simulations show that the two end-effectors can align with the grasping points of the spinning target while maintaining safe clearance to the obstacle by the proposed NMPC method, which delivers better obstacle avoidance distances than the conventional proportional plus derivative (PD) method. The simulation results indicate a practical balance between control performance and computational feasibility for on-orbit operations by dual-arm space robots.