Research On Trajectory Planning Of Intelligent Vehicles Changing Lanes In Foggy Weather

The new generation of automotive industry revolution is in the ascendant,and the "intelligence" of vehicles has become a strategic goal for the development of the global automotive industry.The perception ability and the comprehensiveness of the scene covered by the automatic driving function are the most direct standards for measuring the "intelligence" level of intelligent vehicles.Changing lanes in foggy scenes is a big challenge for both drivers and smart vehicles.In the foggy scene,the intelligent vehicle’s perception system’s wrong judgment of the environment will lead to the generation of dangerous lane changing trajectories.Therefore,this paper studies the lane-changing trajectory planning of intelligent vehicles in foggy scenes.The main research contents of this thesis are as follows:(1)Aiming at the fog level detection problem of smart vehicles,a fog level detection model is established.According to the degree of influence of foggy days,the foggy days are classified into three types: light fog,mist and thick fog.Based on the HSV color space characteristic V/S value,the vanishing point detection method is used to set the calculation area in the image,and the time filter is used to calculate the characteristic value within a certain time range.The experiment collected foggy video data,and determined the judgment threshold in the model by counting the feature values of 3000 frames of micro-fog and dense fog-level images.(2)Improve the dark channel defogging algorithm,improve the image quality in foggy days,and train the vehicle detection model based on the YOLO-v3 target detection network.By quickly estimating the atmospheric light value and the refined transmittance,the two problems of the original algorithm,the long time and the halo appearing in the image after defogging,are effectively solved,and the advantages of this algorithm are verified by comparison with five defogging algorithms.Real vehicles collect driving videos under different foggy weather levels to produce foggy vehicle detection data sets.Combined with the image defogging algorithm,the efficiency of vehicle detection in foggy days is improved.(3)Considering the impact of fog and the safety of vehicles and passenger comfort when changing lanes,a foggy lane changing trajectory planning algorithm is proposed,and the safety of vehicles changing lanes is guaranteed through lane changing safety detection.Based on the fifth-order polynomial,a cluster of candidate trajectories satisfying the constraints of lateral acceleration and maximum vehicle speed is constructed,combined with the classification of foggy days,a trajectory evaluation function considering the influence of foggy days is designed,and the corresponding lane-changing trajectories are selected according to different foggy days.Under the dense fog level,the vehicle has the best safety and comfort;under the light fog level,the vehicle lane change efficiency is the best.A safety distance model considering the influence of fog is established.When safety detection is performed under different fog levels,the safety range of the vehicle model is different.(4)Based on model predictive control(MPC),a lane-changing trajectory tracking controller is designed.Through Car Sim and MATLAB/Simulink co-simulation,the safety and feasibility of the foggy lane-changing trajectory planning algorithm are verified.The simulation results show that the established safety distance model improves the safety of lane changing,and the planned trajectory is safe and feasible for the vehicle.