Research On Autonomous Vehicle Behavior Decision-Making And Trajectory Planning

Autonomous driving is one of the important development directions in the field of modern transportation,which can significantly improve the efficiency of road traffic and ensure the safety of vehicles.Behavioral decision-making and trajectory planning are key technologies for autonomous driving,and their intelligence affects the safety and reliability of vehicle driving,as well as adaptability in complex and changeable real-time traffic environments.This paper focuses on the behavior decision-making and trajectory planning of autonomous vehicles,as follows:In order to ensure the legality and safety of vehicle driving behavior,a behavior decisionmaking method based on optimal utility is designed.By analyzing the driving behavior characteristics of drivers,the design criteria of the behavior decision-making system are clarified.The system analyzes the static traffic information requirements and creates a set of vehicle candidate driving behaviors.The evaluation indicators of driving behavior with different priorities are designed,and the optimal driving behavior decision of the vehicle is determined by calculating the comprehensive utility value.Simulation analysis verifies the effectiveness of the behavioral decision-making method.In order to improve the decision-making success rate of multi-agent reinforcement learning algorithm in autonomous vehicle behavior decision-making tasks,a multi-vehicle behavioral decision-making method based on deep reinforcement learning is studied.Based on the Proximal Policy Optimization(PPO)algorithm,a multi-agent reinforcement learning algorithm IPPO(Independent PPO)using independent learning and parameter sharing is proposed.The MDP(Markov Decision Process)model of autonomous vehicle behavior decision-making was established,and the state space,action space and reward function were designed.Through the simulation experiment of highway ramp scene,the results show that the IPPO algorithm can obtain a higher average reward,and the vehicle can complete the ramp merging task faster and with a higher success rate.In order to improve the adaptability of the lane change trajectory of autonomous vehicles under different driving conditions,a lane change trajectory planning method based on multiobjective optimization is designed.By comparing the commonly used lane change trajectory planning methods,the five-order polynomial planning lane change trajectory was determined.The lane change collision detection algorithm is used to determine the time range of vehicle safe lane change.A multi-objective optimization function that comprehensively considers lane change comfort and lane change efficiency is established,and the weight of the optimization function is dynamically adjusted by the fuzzy controller.The simulation experiment results show that the lane change trajectory planning method can cope with the autonomous lane change of vehicles under different lane changing conditions,and plan the lane change trajectory that adapts to different driving conditions.Through the joint simulation of Prescan and Simulink,the effectiveness of the designed behavioral decision and trajectory planning method is verified.Two typical traffic scenarios are established for simulation experiments,and through the analysis of experimental results and experimental data,the results show that the behavior decision-making method can make the optimal driving behavior for autonomous vehicles,and the decision-making results are in line with the actual situation.The trajectory planning method can dynamically adjust the weight coefficient of the multi-objective optimization function according to the driving conditions of the vehicle,and the planned lane change trajectory is safe and effective,and can adapt to different driving conditions.