Hierarchical Trajectory Optimization Method Based on Dynamic Potential Fields in Complex Scenarios

This paper investigates the trajectory optimization problem for autonomous ground vehicles (AGV) in complex scenarios, considering both dynamic and static obstacles. To describe the problem, a novel lightweight potential function is proposed and an optimal control problem (OCP) is formulated. Due to the high complexity of the problem, conventional numerical trajectory planning algorithms cannot directly solve it. This paper proposes a hierarchical trajectory optimization approach based on dynamic potential fields, consisting of the path planning layer and the trajectory optimization layer. In the first layer, an improved artificial potential field method is employed to solve the initial path. In the second layer, the optimal control problem is transformed into a nonlinear optimization problem (NLP), with the initial path discretized as the initial guess for the numerical solution. The effectiveness and efficiency of the proposed algorithm are verified through numerical simulations.