Aiming at the redundancy characteristics of the redundant manipulator and the relevant RRT^* algorithm had issues with low search efficiency, unstable convergence and not fully considering the influence of the geometric configuration and self-motion characteristics of the redundant manipulator's end effector on path planning in applications for planning the end-effector path, this paper proposed an improved strategy. Firstly, this paper introduced a sampling point selection method based on the angle between the root and end node connection, along with the establishment of a target proximity region. This method dynamically switched between improved sampling methods and random sampling methods based on the number of successful continuous samples. Subsequently, this paper combined the bidirectional tree expansion strategy with the aforementioned method. Finally, it reconnected the initial feasible path twice to obtain the optimized path. Through verification, it improved that the bidirectional tree RRT^* method could effectively enhance search efficiency, convergence, and path optimality. The introduction of virtual collision bodies and capsule collision bodies could also effectively respond to the impact of the structure and motion characteristics of the end effector of the redundant manipulator. This paper used the Mujoco physics simulation engine to validate the movement of the redundant manipulator. This paper demonstrates that this strategy can plan a more optimal feasible end-effector path for the redundant manipulator.

[2023-11-14] Accepted Paper /doc/2023.06.0275/v1

[2024-02-05] Printed Article /doc/2023.06.0275/v2