Improved rapidly-exploring random tree star collaborative path planning considering dynamic redundancy behaviors for dual mobile manipulators

The collaborative transportation of dual mobile manipulators (DMMs) requires path planning for both target objects and mobile bases. Existing studies usually optimize the mobile base path when the target path is predetermined to avoid collisions. However, in many cases, the target path is unknown and must be planned collaboratively with the base path. In addition, DMMs are redundant systems, and their dynamic redundancy behaviors should be considered to achieve flexible obstacle avoidance and better routing. Therefore, this paper proposes an improved RRT*-based collaborative path planning method for DMMs considering dynamic redundancy behaviors. First, a spring formation model is developed based on transportation posture optimization and a capability map. Then, map information is preprocessed through obstacle expansion and obstacle region division. Furthermore, the obstacle avoidance and node reachability judgment of RRT* are improved, and a heuristic cost function is introduced to evaluate path performance. Simulation results show that, compared with RRT*, the proposed method reduces the average cost by over 60% and the average path distance by over 30%. Experiments on a real DMM platform further verify the feasibility of executing the planned target and mobile-base trajectories under the tested representative conditions.