Research On Vehicle Lane Change Prediction And Anti-collision Early Warning Algorithm

In recent years,with increasing number of domestic cars and drivers,traffic accidents have occurred frequently,and among them,lane-change maneuver was a main cause of serious accidents.Therefore,automobile active safety technology has attracted people’s attention.In order to ensure a safety driving,people have developed advanced driver assistance systems(ADAS)for improving the driving safety of vehicles,such as the frontal collision warning system(FCWS),automatic emergency braking(AEB),and lane keeping assist(LKA).Through the research of relevant literature at home and abroad,currently,most lane change prediction methods require a large amount of training data,which usually means a huge economic and time cost to collect the actual data of different cases to verify the algorithm.In addition,with the increase number of target vehicles,the computational efficiency of the algorithm will be greatly reduced,which limits the applicability of these methods in vehicle mounted systems.On the other hand,many approaches rely on the information collected or extracted from radar or video.However,these sensors or systems suffer from some drawbacks,for example,easily affected by obstacles,weather,and other environment factors.In this dissertation,based on the transmission of Basic Safety Messages(BSMs)sent by short distance communication between vehicles(DSRC),considering the realization of efficient and reliable lane change prediction and warning subsystem in ADAS system,in order to improve driving safety.(1)This paper designs a vehicle lane change prediction algorithm based on vehicle status information and driving environment information.The input data of the algorithm comes from the security message set information transmitted between vehicles,which ensures the real-time calculation of the lane change prediction algorithm on the vehicle terminal.The lane change prediction algorithm has a clear design idea which modeling the lane change probability reflected by the vehicle’s motion status,modeling the lane change probability reflected by the traffic environment,and combing the two parts lane change evidence to achieve the comprehensive judgment,so that the algorithm has a strong applicability.Several road scenes are established on the VISSIM simulation software to test the feasibility of the lane change prediction algorithm.The results show that the algorithm can be applied to various lane change scenarios.And the effectiveness of the algorithm was tested using 1,000 highway vehicle data collected by the US Department of Transportation’s NGSIM project.The experimental result is that the accuracy rate of 500 lane-changing data is 95.8%,and the accuracy rate of 500 non-lanechanging data is 88.4%.The vehicle lane change is predicted 0.98 s before the car reaches the boundary line,which proves the effectiveness of the algorithm.(2)On the basis of lane change prediction,considering the factors such as safe distance,danger level and grading warning,we designed and implemented an anticollision early warning algorithm that can be used in the advanced driver assistance system ADAS.In order to avoid the system’s distrust caused by the driver’s frequent interruption,the system alarm level is divided into two types,namely the alarm based on the reminding meaning and the alarm based on the warning meaning.