Trajectory Planning for Intelligent Vehicles in Unstructured Environments: An Unconstrained Optimization Method Based on Improved Hybrid A*

With the advancement of autonomous driving technology, its applications have expanded from urban roads to unstructured environments such as wilderness and disaster areas. While the traditional Hybrid A* algorithm can accomplish trajectory planning, it falls short in optimizing obstacle avoidance distance and path smoothness. To address these limitations, this paper proposes an unconstrained optimization method based on an improved Hybrid A* algorithm. By integrating front-end path point search, linear fitting, and trilinear interpolation, and incorporating an unconstrained optimization algorithm for trajectory refinement, the method significantly enhances the smoothness of speed and acceleration, obstacle avoidance capability, and overall trajectory fluency. Experiments conducted in unstructured environments on the ROS/Rviz simulation platform demonstrate that the proposed approach reduces the standard deviation of the average obstacle avoidance distance by 13.94%, decreases the average trajectory curvature by 16.69%, and lowers the maximum trajectory variation rate by 45.76%, outperforming conventional methods.