Trajectory Optimization for Non-Prehensile Manipulation of Space Robots

Abstract

The increasing demand for on-orbit servicing tasks has driven advancements in space robotics. Traditional capture-based manipulation methods are limited by the need for customized grasping mechanisms and target-specific designs, reducing adaptability for uncooperative or docking-less targets. Non-prehensile manipulation through contact offers a promising alternative for OOS, especially in low-gravity environments. However, the complexity of contact interactions have posed significant challenges to progress in this area. This paper proposes a mode-invariant trajectory optimization method for non-prehensile manipulation. The proposed approach combines complementarity constraints with a direct trajectory optimization framework and employs sequential quadratic programming solvers to generate contact-aware trajectories for non-prehensile manipulations. Simulations involving a 7-degree-of-freedom space robot reorienting a target object demonstrate the method’s feasibility and effectiveness. The results show the capability to produce physically realistic and efficient trajectories, enabling complex tasks without relying on capturing mechanisms.