Recognition Metood Of Lane-changing Behaviour And Danger Level For Intelligent Vehicle

With the increase of traffic accidents year by year,the active safety technology of intelligent vehicle has gradually become the focus of people's attention.Lane-changing behavior and danger level recognition is an important part of active security technology,which can efficiently avoid the collision of vehicles and decrease traffic accidents.In this paper,the recognition method of a front vehicle's lane-changing behavior and danger level in the adjacent lane is studied,in order to identify the lane change accurately and quickly,and take measures on avoiding collision according to specific circumstances.The research in this paper has a great meaning on improving the safety of driving.In this paper,the lane change behavior and danger level recognition method of intelligent vehicle is studied in four chapters.A statistical model of vehicle lane changing recognition process is established,and the judgment method of lane-changing behavior and danger level is designed;The parameter training algorithm of the model is derived,and a simulation environment is constructed;The validity of the algorithm is analyzed in the simulation environment.In this paper,hidden Markov model is used to establish lane change identification model according to the characteristics of lane change identification process.The safe lane-changing process and the dangerous lane-changing process are defined according to the relative danger level of the front vehicle's lane change to my car.According to the features of lane keeping and lane changing,and features of safe lane change and dangerous lane change,lateral distance between two vehicles,lateral speed of front vehicle,longitudinal distance between two vehicles and my vehicle's speed are chosen as observation variables.The hidden states are divided into lane keeping,lane changing state 1 and lane changing state 2,so the safe lane change identification model and dangerous lane change identification model are established.The identification method of sliding time window is designed,and the driving state of each time window is judged in turn,and the method of judgment is given.In order to determine the parameters of the lane change recognition hidden Markov model,a parameter training algorithm is derived.Based on the veDYNA vehicle dynamics simulation software,the simulation environment is built and the driving mode is designed.Data are obtained from the simulation environment and processed after then,so the training samples are obtained.By writing the parametertraining codes in Matlab software,the parameters of the safe lane-changing recognition model and the dangerous lane-changing recognition model are obtained.In order to test the recognition effect of the lane-changing behavior and danger level recognition algorithm,the test samples are obtained from the simulation environment,including the safe lane-changing samples and the dangerous lane-changing samples.Based on the identification of the sample,the recognition process of sliding time window is described,and the state judgment of each time window is proved to be quite accurate.The test results in the simulation environment show that,the recognition rate of the algorithm for both safe lane change and dangerous lane change is rather high.The algorithm is able to make recognition before the front vehicle crosses the line.