| During the driving process,there are countless traffic accidents caused by improper lane change,especially on expressways,where vehicles are driving at a high speed.Rapidly changing lanes may lead to rear-end collision of the vehicles in the target lane,resulting in immeasurable consequences.In order to improve the safety of lane-changing process,this paper analyzes whether the driver has the intention of lane-changing,and carries out vehicle trajectory planning and path tracking according to the safety evaluation indexes among vehicles in the surrounding environment.Firstly,the parameters of driver’s intention recognition are selected.CarSim and Simulink are used to build a vehicle lane-changing model,and the changes of various parameters in the lane-changing process are collected.After screening,analyzing and comparing the collected data,the steering wheel angle signal and the vehicle lateral acceleration signal are selected as the characterization parameters of the driver’s intention to change lanes.Secondly,the method of driver intention recognition is studied.This paper introduces the relationship between hidden Markov model and driver’s intention,and expounds three problems of hidden Markov model and K nearest neighbor classification method.Using the position relationship of vehicles in NGSIM data set combined with hidden Markov model,the state probability matrix and state transition matrix of left lane change,straight lane change and right lane change are obtained through training.The collected signals are made into data sets,and the KNN algorithm is used to classify,train and test the data.The fuzzy matrix shows that the recognition rate of intent is as high as 83.4%,and the recognition accuracy can be improved by increasing the number of data sets or adjusting the ratio of training set to testing set.Then,plan the lane change trajectory and track it.The quintic polynomial is used to fit the lane-changing trajectory cluster,according to the speed comparison between the front and rear vehicles and the self-vehicle,the size of the safe distance between vehicles is calculated.By analyzing the coexistence of peripheral vehicles,the safety evaluation index is calculated,and the optimal lane-changing trajectory is determined at4.4 seconds lane-changing time.Using the combination of CarSim and Simulink,the optimal path is set as the reference path,MPC is used as the controller to track the trajectory,and the maximum error is less than 0.08 m,which has good tracking accuracy.Finally,the hardware-in-the-loop test scheme is designed.Connect the hardware device with the NI acquisition card,first collect the steering wheel angle signal,steering wheel angle speed and accelerator pedal signal,then debug the servo motor with Panaterm and Lab VIEW software.After the combination of CarSim and LabVIEWRT system,build a simulated driving scene to simulate the process of vehicle lane changing,and then collect the steering wheel angle signal in this process and bring it into KNN algorithm for verification,and the recognition accuracy rate is85%.The results show that it is feasible to use the steering wheel angle as the main characterization parameter of driver’s intention recognition,which provides an effective idea for future research in the field of intelligent vehicles.Figure 56 Table 7 Reference 83... |
PDF Full Text Request