A Hierarchical Hybrid Framework for Dynamic Obstacle Avoidance of 6-DOF Robotic Manipulators

Abstract

This paper proposes a hierarchical hybrid planning and control framework for safe and smooth obstacle avoidance of a 6-DOF robotic manipulator in static and dynamic environments. To address the limitations of conventional sampling-based planners, an Rapidly-exploring Random Tree (RRT) planning method combined with an artificial potential field (APF) waypoint optimization strategy is developed to improve trajectory smoothness and execution quality. The optimized offline trajectory serves as a reliable reference for online motion adjustment, reducing unnecessary path oscillations while preserving collision-free feasibility. For dynamic obstacle avoidance, a hierarchical control scheme integrating APF, APF-MPC and MPC is proposed, where different controllers are adaptively activated according to the real-time obstacle clearance. The APF controller provides fast reactive obstacle avoidance, while the MPC-based modules perform predictive trajectory optimization under increased collision risks. Simulation experiments conducted on the PUMA560 manipulator demonstrate that the proposed framework achieves safe avoidance of dynamic obstacles while maintaining path quality. The results verify the effectiveness of the proposed hierarchical strategy for efficient and reliable robotic manipulation in complex environments.