Microscopic Vehicle Trajectory Reconstruction Considering Driving Heterogeneity

In recent years,with the continuous growth of car ownership in China,more and more urban roads are congested.How to carry out road traffic control reasonably and alleviate congestion has attracted extensive attention from all over the world.The formulation of reasonable traffic control measures relies on complete and accurate traffic information,among which vehicle trajectory is a valuable traffic information,which is widely used in queue length,traffic flow,real-time traffic state estimation and other aspects.With the rapid development of intelligent connected technology,there are more and more intelligent connected cars on the road.For a long time in the future,there will be mixed traffic flow composed of artificial driving vehicles and intelligent connected vehicles.Although the intelligent connected vehicle can provide the trajectory information of its own and some of the surrounding vehicles,it is still difficult to obtain the trajectory data of the full sample vehicle in the intelligent connected environment because the artificially driven vehicle cannot provide its own trajectory.How to reconstruct the complete trajectory with high precision from partial vehicle trajectory has become a hot issue to be solved urgently.Therefore,on the basis of considering driving heterogeneity,this paper makes an in-depth study on the micro-vehicle trajectory reconstruction method.Since this study is mainly based on vehicle following characteristics,firstly,four common vehicle following model formulas,characteristics and model parameter calibration algorithms are briefly introduced in this paper.Then,from the perspective of driving external heterogeneity,the selection of characterization indexes of driving external heterogeneity was briefly analyzed.Then,33 tracks were selected from NGSIM I-80 dataset for small sample test analysis.According to the small sample test results,characterization indexes were set.Based on the complete trajectory reconstruction method proposed by predecessors for sporadic and fragmentary tracks detected by intelligent vehicles,and considering the influence of external driving heterogeneity(multiple following models),the trajectory reconstruction method was verified and analyzed using NGSIM data.Then,from the perspective of driving internal heterogeneity,the definition of driving internal heterogeneity is briefly introduced.Then,some tracks are selected from NGSIM I-80 data set to conduct a branch driving state reconstruction experiment,and the experimental results are analyzed to explore whether drivers show obvious driving internal heterogeneity under different driving states.And how to consider the influence of driving internal heterogeneity on the optimization reconstruction process to improve the trajectory reconstruction accuracy.Finally,on the basis of the previous research results,focus on the reconstruction of F(following)state trajectory,and improve the existing trajectory reconstruction method considering the influence of internal driving heterogeneity.In order to verify the rationality of the improved method,part of NGSIM data is used to verify and analyze the method.The experimental results show that driving internal heterogeneity has a greater impact on actual traffic than driving external heterogeneity.In trajectory reconstruction,the trajectory calibration and tracking model with the same driving state as the reconstructed trajectory should be selected as far as possible to reduce the trajectory reconstruction error and improve the reconstruction accuracy.At the same time,compared with the reconstruction of other driving state tracks,the reconstruction of F(following)state track has the largest error,which should be paid more attention to.