Study on Trajectory Planning Method Using Adaptive Weight-Based Variable Interval Sampling Strategy

The study addresses the issues of slow real-time trajectory planning and poor passenger comfort observed in autonomous vehicles operating in complex road environments with static and dynamic obstacles. A trajectory planning method based on an adaptive weight strategy with variable sampling intervals is proposed. First, a variable interval sampling model is developed to address the discomfort caused by the constant sampling intervals of the fifth-order polynomial in the traditional Frenet coordinate system, particularly in situations with varying curvatures. Next, to address the issue of slow trajectory planning speed, an adaptive weight loss function model is constructed, considering conditions such as kinematic feasibility, boundary state constraints, and collision avoidance. Finally, the Red-billed blue magpie optimizer is used to solve the adaptive weight model in real-time, further improving planning speed. Simulation results show that the proposed method improves the overall path planning speed by an average of 8.86% and enhances the com-fort index by an average of 13.14% compared to traditional algorithms.