Research On Path Planning And Tracking Control Method Of Campus Intelligent Vehicle Based On Hybrid Algorithm

Thanks to the support of big data and intelligent networking technology,intelligent driving vehicle has gradually become a hot research topic in the automotive field.Compared with traditional vehicles,intelligent vehicles can achieve safer and more reliable driving output through intelligent driving system,further improve occupant comfort,better ensure the safety of human life and property,and further alleviate the current situation of traffic congestion and energy shortage.As an important part of intelligent driving system,path planning and tracking control algorithm is the basis for vehicle to obtain collision free safe path and accurate tracking.Based on the mastery of environmental information,path planning can be divided into global path planning and local path planning.Through the analysis of the research status of different algorithms,it can be found that the global path planning algorithm generally has many defects,such as many twists and turns of the planned path,close to the road boundary and the exponential growth trend of calculation time with the increase of map scale;The local path planning algorithm has the defects of insufficient real-time,easy to fall into local optimization and insufficient consideration of vehicle kinematics constraints;In the aspect of tracking control,there are some defects,such as the lack of tracking accuracy,and the output of control quantity may not meet the constraints of vehicle dynamics or kinematics.Therefore,this paper takes the intelligent vehicle driving in the campus environment as the research object and the path planning and tracking control algorithm as the research core.The main work is as follows:(1)This paper studies the global path planning algorithm of intelligent driving vehicle,and analyzes the defects of the traditional A* algorithm,such as many twists and turns of the planned path,close to the road edge and the exponential growth trend of calculation time with the increase of the scale of grid map.Therefore,from the perspective of improving the calculation efficiency of the traditional A* algorithm,the safety and feasibility of the planned path,the improved A* algorithm is proposed in this paper by introducing the map preview module,the collision field model based on the safe distance,improving the evaluation function and the smoothing method of quasi-uniform cubic B-spline curve.Simulation experiments on different types of scene maps show that the improved A* algorithm improves the computational efficiency of the traditional A* algorithm,and significantly improves the security and feasibility of the planned path on the basis of similar path length to the traditional A* algorithm.(2)This paper studies the local path planning algorithm of intelligent driving vehicle,and compares the dynamic windows approach(DWA)algorithm and vector field histogram(VFH)algorithm,which belong to the local path planning algorithm generated based on path cluster,in the same environment map.Through comparative experiments,it is found that DWA algorithm has higher smoothness and feasibility because it updates the state based on the kinematic model in the planning process and considers the dynamic range of speed change in the speed sampling process;The VFH algorithm only considers the feasible path based on obstacle information,and the path has many twists and turns and poor real-time performance.Therefore,this paper finally selects the DWA algorithm to be applied to the local path planning link.At the same time,in order to further meet the needs of intelligent driving tasks,the DWA algorithm is improved.The evaluation function of the traditional DWA algorithm is improved by integrating the global planning path information and introducing the angular velocity evaluation factor.Finally,in order to further determine the weight setting of different evaluation factors,The multi-objective grasshopper optimization algorithm(MOGOA)is used to optimize the combination weight.The simulation results show that the improved algorithm effectively improves the comprehensive performance of the traditional DWA algorithm.(3)The path tracking control algorithm of intelligent driving vehicle is studied.In order to ensure the tracking accuracy of the vehicle and consider the requirements of vehicle kinematic constraints and dynamic constraints,model predictive control(MPC)is used to build the path tracking controller.Firstly,the kinematics model of the whole vehicle is built based on the main parameters of the campus intelligent vehicle.The linear error model of the vehicle is obtained by Taylor expansion at the reference point.The error model is discretized by the forward Euler discretization method,and then the final prediction model of the path tracking controller is obtained.Finally,the objective function of MPC algorithm is designed based on the actual task requirements,and the control quantity and control increment range are set based on the vehicle kinematics constraints.The simulation results show that the path tracking controller based on MPC can realize high-precision tracking within a certain error range.At the same time,its output control quantity is always within a certain constraint range,and the control increment does not change dramatically,which meets the requirements of vehicle stability.In order to realize the speed tracking control of intelligent vehicle,the Stanley algorithm is used to build the speed tracking controller.The performance of the speed tracking controller is further tested through the steady-state speed condition and step speed condition.The simulation results show that the speed tracking controller can realize the stable tracking of the expected speed and ensure the smooth and reliable output,which has certain practicability.(4)The overall system of path planning and tracking controller is tested through joint simulation to further verify the performance of the hybrid algorithm.