Evaluation of false alarm alarms in truck FCW based on calibration of RSS model under different driving scenarios  

作　　者：Yanli Bao Xuesong Wang Rongjie Yu 

机构地区：[1]College of Transportation Engineering,Tongji University,Shanghai 201804,China [2]Key Laboratory of Road and Traffic Engineering,Ministry of Education,Shanghai 201804,China

出　　处：《International Journal of Transportation Science and Technology》2024年第3期35-50,共16页交通科学与技术(英文)

基　　金：sponsored by the National Key R&D Program of China(2018YFB0105202).

摘　　要：Advanced driver-assistance systems(ADASs),such as forward collision warning(FCW),are widely used and,in some countries,have been made mandatory for commercial vehicles.In practical applications,however,FCW systems produce many false alarms.Using scenario and driving behavior data collected from naturalistic driving study data of trucks,a variable threshold evaluation method was proposed to determine the factors correlating with false alarms.A total of 450 collision avoidance events were divided based on driving character-istics into three groups with k-means clustering.Responsibility-sensitive safety(RSS)mod-el’s parameters were calibrated with the driving behavior characteristics and scenarios to evaluate the truck FCW system’s alarm accuracy.The evaluation of the results of truck FCW system based on RSS model found 47 false alarm alarms in the 450 events,a false alarm rate of 11.19%.When the following distance was close(<7 m)or far(>20 m),the false alarm rate reached more than 30%.The minimum time to collision(TTC)in the close distance driving clusters(DCs)(5.81 s)was lower than that in long distance DCs(7.68 s and 9.46 s).Braking force in the low-speed DCs(deceleration at0.16 g and0.55 g)was lower than in high-speeded DC(deceleration=1.21 g).The FCW system does not conform to the driver’s reaction time and braking characteristics in different scenarios,and is the main reason for false alarms.This is more obviously reflected in low-speed short distance and high-speed long-distance scenarios.

关 键 词：Truck forward collision warning Risk evaluation Active safety system data Responsibility-sensitive safety model CALIBRATION False alert analysis 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]