Typical Test Scene Extraction And Scene Library Construction Based On ADAS HIL System

With the development of automobile technology,traffic safety has always been the focus of people's attention.In recent years,although the total number of traffic accidents in China shows a downward trend,it is undeniable that the traffic safety problem facing China is still quite serious.In order to reduce the probability of accidents and further improve the driving safety,the active safety technology of automobile is attracting extensive attention of researchers.Based on this,advanced driver assistant system(ADAS)products represented by lane departure warning(LDW),autonomous emergency break(AEB),front collision warning(FCW)have developed rapidly in recent years.However,before putting these products into production,a large amount of testing and verification needs to be carried out in various test scenarios,which are difficult to carry out in the actual environment.Therefore,in order to verify these functions,the establishment of typical virtual test simulation scene is a very important part in the early stage of simulation test.The test scenario built in this paper is typical traffic hazard scenarios,providing a test environment for ADAS functions.First of all,by studying the road traffic accidents in tianjin in 2017,this paper makes a careful analysis from the following aspects: weather,time,lighting,road conditions,traffic control methods and accident forms to extract the main influencing factors at the time of the accident.Based on these main influencing factors,it is used as a reference for selecting scene elements when constructing a typical dangerous scene.This paper uses tianjin taxies equipped with millimeter-wave radar,dashcam and data acquisition terminal as the data acquisition platform,capturing video conditions of vehicles driving in natural driving conditions.By viewing the video playback,the video was converted into data and recorded.A total of 456vehicle-motor vehicle,vehicle-non-motor vehicle,and vehicle-pedestrian dangerous conditions were obtained.The system clustering method was used to combine the similar dangerous working conditions.finally,six types of vehicle-motor vehicle typical dangerous working conditions,five types of vehicle-non-motor vehicle typical dangerous working conditions and four types of vehicle-pedestrian typical dangerous working conditions are obtained.Secondly,this paper uses CarMaker software which is a part of ADAS HIL system to set the scene elements according to the scene elements in typical dangerous working conditions,building virtual simulation scenes with video animation effects,forming a library of typical dangerous scenes for vehicles-motor vehicles,vehicles-non-motor vehicles and vehicles-pedestrians,which is used as the development and test environment of ADAS algorithmFinally,based on the typical dangerous scenario of vehicles-people,for example,the paperjointed CarMaker and Simulink simulation to test automatic emergency braking system function,and verified the reliability of the laboratory existing control algorithm;In addition,based on the ADAS HIL system and MicroAutoBox,combined with the existing perception,decision planning and control algorithms in the laboratory,the collision avoidance control test was conducted for the typical vehicle-pedestrian dangerous working conditions.The perception algorithm in the sensor is used to identify pedestrians in the dangerous scene and predict the motion trajectory.The B-like curve algorithm is used to generate a smooth collision avoidance path with time,speed,and direction.Longitudinal control,real-time tracking planned collision avoidance path,thus avoiding collision.