Study on Vehicle Interaction Strength on Urban Road Based on Energy Field
Due to the complexity of traffic environment and the interaction among vehicles,there are some bottlenecks about poor application flexibility and low precision in the study on autonomous driving decision-making behavior.Therefore,taking interaction exploration among vehicles as an object,in-depth studying the micro-law of vehicle flow operation,and provideding theoretical basis for the study on autonomous driving decision-making behavior.Firstly,based on the physical properties of the energy field,such as speed,acceleration and car type,the vehicle is analogized to the field source in the gravitational field,and the vehicle interaction field model is established.The interaction among vehicles is quantified by using the vehicle-vehicle interaction model,and the distance and speed in traffic environment are redefined according to the critical space headway and Doppler effect.Secondly,the data of speed and relative distance are collected by the real vehicle on Tongshun boulevard in Shanghai Pudong,and the data in the event data recorder are extracted by video processing software Kinovea,and Gauss mixture model is used to test the relationship between driving interaction and actual driving behavior.Finally,the driving interaction force is analyzed by K-means algorithm,and the driving risk grade is quantified.The result shows that(1)the errors between the car-following and lane-changing results calculated by Gaussian mixture model and the actual results are 1.12%and 9.1%respectively,indicating that the driving interaction force has a good ability to describe the vehicle interaction quantitatively;(2)at the same time,the driving risk is divided into 4 levels based on the driving interaction force model,and the driving risk is effectively evaluated.The proposed driving interaction force model can not only expand the application of previous driving safety field,but also provide theoretical support for driving behavior decision-making study.
ITSdriving interactionenergy fielddriving riskcritical space headwayGaussian mixture model
NETL
NSTL
万方数据
