Moving Horizon Obstacle Avoidance Path Planning Of Intelligent Vehicle Based On Driving Style

In recent years,with the rapid development of intelligent vehicle,path planning as one of its key technologies has been widely studied.Route planning needs to consider environmental information such as vehicle state,driving state and traffic signal state to determine the optimal driving route,which directly determines the driver's driving comfort and driving safety.Drivers have similar requirements for safety,but different drivers have different requirements for comfort.Considering the appropriate driving style is an important measure of driver comfort.How to provide personalized path planning to adapt to the driver's driving style is the focus of this paper.This paper is supported by the National Natural Science Foundation of China project 3(No.61790563)"Human-machine Cooperation Mechanism and Switching Control under Extreme Operating Conditions",National Natural Science Foundation of China(No.U1664263)"Dynamics modeling and cooperative control method of driverautomation copiloted intelligent vehicles" and National Natural Science Foundation International(Regional)Cooperation and Exchange Key Project(No.61520106008)"SafetyOriented Energy Efficiency Moving Horizon Optimization for Electrified Vehicles".The research on path planning mainly includes the following aspects:In order to plan personalized driving paths based on driving style,driving style is first identified and divided into "radical","smooth" and "conservative".A wide range of drivers are selected to conduct simulator driving experiments to obtain vehicle status information.The key features of driving style representation state extraction are analyzed,and all drivers are classified into three categories based on fuzzy rules.Considering that speed is the state that reflects driving style,a speed factor is proposed by synthesizing the speed of each driver and the maximum allowable speed as the input of the subsequent path planning.Due to the complexity of the actual road environment,with the continuous change of the vehicle driving environment,the cost and time required for real-time environmental detection and modeling depend entirely on multi-sensor are high.In view of the dynamic change of environmental information,this paper divides the lane area and analyses the movement law of traffic components such as roads and obstacle vehicles.Based on the virtual force field environment description method,a simplified environment model including the virtual gravity field of the desired driving target,the dynamic rectangular virtual repulsion field of obstacle vehicle and the virtual gravity field of the lane area is established,and the traffic environment is described with fewer parameters.In view of the difficulty of considering vehicle dynamics in the planning stage in the way of directly planning the traveling route and then tracking the path,this paper establishes vehicle kinematics and dynamics model based on model prediction method,transforms the path planning problem into optimization problem,considers environmental information and dynamic constraints,and directly optimizes vehicle state information as a reference for tracking control.Firstly,vehicle stability constraints are transformed into constraints in model predictive control.Then,according to the established virtual potential field model,three optimization objectives are abstracted: Lane maintenance,obstacle avoidance and target orientation.According to the identified speed factor,the planning speed optimization goal 4 is provided.According to the vehicle dynamics characteristics,the optimization goal 5 of smoothing the change of control state is proposed.By using the weight factor,the multi-objective optimization problem is transformed into a single-objective optimization problem.The simulation results show that the route planning results of different driving styles can be provided under the condition of guaranteeing smooth and safe obstacle avoidance.Because the planned controller is designed to plan the required vehicle state directly from the environmental information,the change of the environment will lead to the change of the planned vehicle state,and the change of the state is particularly sensitive due to the shortening of the length of the planned and controlled channel,resulting in the phenomenon of frequent state jitter.Based on the control period and model accuracy,the multi-scale controller and dynamic model are improved.The simulation results show that the improved scheme reduces the vehicle state jitter and optimizes the control input more smoothly.