Research on Local Path Planning Algorithm for Autonomous Vehicles Based on Polynomial Curves

With the rapid development of artificial intelligence technology, autonomous driving technology has been widely applied in smart transportation, logistics, agriculture, and other fields. At the same time, as traffic safety and environmental issues become increasingly prominent, the emergence of autonomous driving technology can effectively address traffic safety problems caused by human factors and improve environmental issues, playing a significant role in promoting social progress and technological development. This paper proposes a local trajectory planning method based on quintic polynomial curves in the Frenet coordinate system, decoupling the trajectory planning problem into lateral and longitudinal motion planning problems. First, sampling is conducted separately for lateral and longitudinal movements to generate sets of lateral and longitudinal trajectories, and trajectory evaluation functions are designed from the perspectives of comfort, safety, and other factors. Then, the lateral and longitudinal trajectories are synthesized, and a comprehensive evaluation function is designed to assess the synthesized trajectory. Finally, the feasibility of the proposed local trajectory planning algorithm is verified through simulation tests under conditions of static obstacles on curves and dynamic obstacles on straight paths. The results show that the path planned by this algorithm has good smoothness, can reach the target speed during lane changes, and exhibits gradual speed changes during acceleration or deceleration without abrupt speed variations, meeting the driving requirements of vehicles in real-world scenarios.