Research On Local Trajectory Planning And Tracking Control Algorithms Of Intelligent Vehicles

The front end of the trajectory planning is connected to the perception system,and the back end is connected to the control system,which is one of the functional cores of the entire unmanned driving system.The trajectory tracking control system inputs control instructions for the smart car’s actuator,which is the basis for the smart car to realize unmanned driving.However,with the commercialization of autonomous driving technology and the complexity of driving scenarios,the effectiveness,real-time and safety requirements of trajectory planning and control algorithms are higher.Therefore,this paper mainly studies the local trajectory planning and tracking control algorithm of intelligent vehicles in multiple scenarios.The specific research contents are as follows.Firstly,the Frenet coordinate system is introduced,and the horizontal and vertical trajectory planning model of the smart car is established based on the Frenet coordinate system.For the common constant speed cruise,lane change and overtaking and deceleration parking lot scenes,the respective terminal sampling states are established,and the trajectory quality evaluation function is designed,and the horizontal and vertical sampling trajectory family with quality evaluation scores is obtained.The horizontal and vertical sampling trajectory synthesis algorithm is proposed,and a new quality evaluation function is designed for the synthesized trajectory.Kinematics constraint detection is introduced to the trajectory set,to filter out trajectories that do not meet the kinematic constraints,oriented bounding box is used to simplify the shape of environmental obstacles and unmanned vehicles,and the separation axis theory is introduced to perform the trajectory Impact checking.Finally,a program verification is designed for common driving scenarios,and the results show the effectiveness,safety and real-time performance of the trajectory planner proposed in this paper.Based on the vehicle dynamics,a model predictive controller is established.For the two parts of path tracking and longitudinal speed tracking,the state equation is established with the horizontal and vertical errors as the state variables.A comparative experiment between the model predictive controller and the pure tracking controller is designed.The experimental results show that the designed model predictive controller has a better tracking effect than the pure tracking controller.Finally,the particle swarm algorithm is introduced to optimize the weight parameters in the model predictive controller,and a comparison experiment between optimization and pre-optimization is designed.The experimental results prove that the tracking accuracy of the model predictive controller optimized based on the particle swarm algorithm is Better promotion.Aiming at the improved model predictive controller,a joint simulation experiment is designed under the double-shifting condition.The experimental results show that the improved trajectory tracker has a good tracking effect on the desired trajectory.The joint decision-making module,trajectory planning module and control module respectively designed joint simulation experiments for static obstacle scenes and dynamic obstacle scenes.The experimental results show that the active obstacle avoidance system designed in this paper can be combined with the decision-making layer to plan smooth parts Obstacle avoidance trajectory,the model predictive controller can quickly track the trajectory,the tracking accuracy is large,and a good active obstacle avoidance effect is achieved.