A Game-Based Fault-Tolerant Path Planning Algorithm For Space Manipulator

This paper studies the fault-tolerant path planning for the space manipulator with single joint locked failure. To search a reasonable path in both Cartesian and joint spaces, a Game-based RRT* algorithm is proposed by integrating information, including both minimum path and minimum singular value, from two spaces. First, the degraded working space for single joint locked failure is established by constructing orientation reachability to the reachable position space. Secondly, to improve the convergence rate of RRT*, the rasterized grid of the degraded working space is used to construct the path tree. Then, integrating the minimum singular value and minimum path to each grid, a numerical steer mechanism based on game theory is designed to find the new grid for a tree. In this mechanism, game theory is introduced to avoid the single parameter dominant caused by different parameter dimensions. Finally, the fault-tolerant path is planned by this improved RRT* algorithm. Simulation results demonstrate that the proposed Game-based RRT* algorithm can search a path that satisfies the requirements of singularity avoidance and path minimum simultaneously.