Neural network-based time optimal trajectory planning method for rotary cranes with obstacle avoidance

Abstract

In this paper, an improved trajectory planning method is proposed, which can effectively avoid obstacles on the path and control the swing of the load during the transportation of a rotating crane within a certain range. The complex dynamics model of the rotary crane is first analyzed, and then the system constraint equations are established to ensure the transient performance during the boom motion. And an optimal time trajectory is obtained by the dichotomy method. Then, the trajectories for different obstacles, different starting positions and different final positions are generated by combining with a neural network method to achieve different control requirements. Finally, to confirm the superiority of the proposed method, it is verified by a hardware experimental platform.