Trajectory Planning for Autonomous Vehicles Based on Interactive Risk Perception and Prediction

This study introduces a trajectory planning framework for autonomous vehicles utilizing the Frenet coordinate, highlighting on maneuverability and safety in urban environments. The framework incorporates an improved A * algorithm for velocity curve planning and a lateral trajectory sampling method. An interactive risk perception model abstracts the vehicle's trajectory into adjustable frames and dynamically defines the risk area to facilitate real-time collision avoidance. This method allows for adaptive trajectory planning by incorporating a predictive interaction model that anticipates the movements of other traffic participants. Simulation results demonstrate the effectiveness of the framework in real-time scenario simulations, highlighting its potential to improve the safety and efficiency of autonomous vehicles navigation in urban settings.