Generate optimal grasping trajectories to the end-effector using an improved genetic algorithm

Determining how to grasp a non-cooperative target by applying a space robot is still a challenging issue, especially when the target is tumbling in space. This paper presents an optimal grasp planning strategy for a kinematically redundant space manipulator to capture an arbitrarily rotating target, such as a dysfunctional satellite or a piece of space debris. The kinematics and dynamics of the non-cooperative target and the space robotic system are first introduced, which lays the foundation for constructing the grasp planning strategy. Subsequently, an optimal grasping time and the target's terminal motion states are determined with consideration of the robotic capability, the target's motion and the grasping direction. Furthermore, the joint trajectories are parametrized with the quintic Bézier curve, and the direct kinematics equations are employed to overcome the dynamics singularity issue. The grasp planner of the space robot is then transformed into a multi-constraint, multi-objective nonlinear optimization problem solved by an improved real-coded genetic algorithm (RGA). The proposed grasp planning scheme is immune to the singularity issue and generates an optimal robotic configuration for post-capture manipulation. Simulation results are presented for capturing a tumbling target using a seven-degree-of-freedom kinematically redundant manipulator mounted on a free-floating spacecraft, demonstrating the effectiveness of the proposed method.