Consistency Evaluation And Optimization Of Dynamic Trajectory Planning For Autonomous Ground Vehicles In The Scene With Moving Obstacles

Combined with computer technology,map positioning,image processing and other information technologies,intelligent vehicles using on-board sensors to perceive the surrounding environment,obtain road and environmental information.They can automaticly control the speed and direction through autonomous decision-making and trajectory planning.This paper focuses on dynamic local trajectory planning in the process of vehicle intelligent driving,aiming at the problem found in simulation and real vehicle testing,that is,the local trajectory planning based on polynomial optimization solution has poor trajectory consistency in obstacle avoidance process,and carries out research on trajectory consistency optimization in obstacle scenarios.With specific contents as follows:(1)After the coordinate conversion between Frenet coordinate system and Cartesian coordinate system is completed,the vehicle and environment information are mapped to Frenet coordinate system.After that we need to work out the final state sampling of the lateral and longitudinal trajectory of the intelligent vehicle.Then the quintic polynomial programming algorithm model is used to efficiently solve the lateral and longitudinal coupling trajectory.The lateral and longitudinal trajectory quality evaluation models and coupled trajectory collision risk cost models are established to obtain the total generation value of the trajectory at the same time.Finally we conduct trajectory enforceability and collision detection to select the optimal trajectory with the lowest total cost(2)In order to solve the problem of poor trajectory consistency in local trajectory planning,we proposed trajectory update model which considers trajectory consistency.The updated trajectory is divided into two parts: splicing trajectory and rigid planning trajectory,so as to ensure the time consistency of the overall obstacle avoidance trajectory by splicing trajectory.The consistency cost function is added to the rigid planning trajectory to constrain the consistency of the driving state of the rigid planning trajectory.Then the MATLAB was used to build dynamic and static obstacle scenes to verify the feasibility of the proposed trajectory update model.(3)To verify the proposed trajectory update model furtherly preparing for the subsequent real vehicle test,we designed the Prescan-Simulink-Carsim co-simulation platform,which is used to build a variety of vehicle driving scenes in Prescan.Combined with the CarSim vehicle dynamics model and control algorithm,we tested the trajectory update model in different scenes.The simulation results showed that after adding proposed trajectory update model,the overall improvement of trajectory consistency in each cycle could guarantee the effectiveness of obstacle avoidance trajectory,which improves the stability and safety of vehicle control.As a consequence the whole obstacle avoidance process of vehicle driving state is more stable.(4)The proposed trajectory update model were written into the intelligent vehicle trajectory planning module by using C++ language,which is ready for testing the algorithm in the obstacle scenes of real life.The test results showed that the overall consistency of the planning results of the proposed trajectory update model was higher,and the trajectory continuity in each cycle of the obstacle avoidance stage is nice and there is no drastic changing phenomenon of trajectory.The vehicle condition is stable which is good for getting more comfortable driving sensation.In addition,the safety distance from the obstacles is more reasonable,which enhances the vehicle’s active obstacle avoidance ability.