Research On Safety Lane Change Warning Method Based On Potential Angle Collision Point

With the increase in car ownership,lane change behaviors caused by self-driving travel are increasing.The degree of danger caused by the lane change behavior has also increased.However,with the application of sensing technology,it is possible to reduce traffic accidents caused by driver misjudgment during lane change.The corresponding parameters are detected through sensors,and the driver is assisted to adjust the vehicle state in time through the onboard device to improve the lane changing safety and the lane changing efficiency.Firstly,this paper constructs a lane change vehicle recognition algorithm based on PCASVM by selecting vehicle explicit motion parameters such as shaft steering angle,distance between steering wheel and lane line,longitudinal acceleration,lateral acceleration,lateral speed,and longitudinal speed.Then,in the process of identifying lane change vehicles,the penalty coefficient C and radial basis kernel function G in SVM are optimized by combining with grid optimization.Finally,1092 sample data were selected and processed from the vehicle driving data of the Berkeley Expressway,and the algorithm proposed in the paper was trained and tested.During the test,the model recognition rate of lane change vehicles reached 98.17%,providing technical support for the establishment of data sharing between lane change vehicles and target lane obstacle vehicles.In order to ensure safe lane change and avoid traffic accidents,an effective lane change assist system is required.In a lane change assist system,it is very important to obtain following elements in time,such as actual distance between vehicles,minimum safety distance and warning signal.To this end,this paper analyzed four kinds of lane change angle collision scenes.Initial position,initial velocity,acceleration,heading angle and kinematics of vehicles were used to calculate the position of potential angle collision points between lane change vehicle and obstacle vehicles.Then,actual distance model was constructed based on potential angular collision points.The minimum safety distance model was also established under the two most unfavorable conditions.In order to achieve the lane change warning,three early warning rules were formulated.We verified the validity of models and early warning rules using vehicle driving video data of Interstate 80 in California.Models and early warning rules constructed in our research can be applied to the advanced active safety systems of vehicle,such as vehicle lane change assist system and active collision early warning system,which can improve the active safety and reduce traffic accidents.