Research On Fatigue Driving Detection Method Of Freight Vehicles

Fatigue driving will affect the driver’s reaction ability,execution ability and judgment ability,which is extremely harmful to road safety.Whether in developing or developed countries,freight vehicle accidents account for more than 10% of the total fatalities in traffic accidents.Among them,due to the professional characteristics of drivers,fatigue driving is one of the main causes of freight vehicle traffic accidents.For freight vehicle drivers,even if they realize that they are in a state of fatigue and are forced by life pressure,they will choose to continue driving to ensure the transportation task in most cases.Therefore,it is of great significance to study the fatigue driving detection of freight vehicle drivers.In the existing studies,there are few studies on the fatigue state of freight vehicle drivers.In order to reflect the driver state more objectively and truthfully,this paper collects a large amount of vehicle operating state data and driver’s facial video data under natural driving conditions,and designs a fatigue driving detection algorithm based on facial features and vehicle running data to complete the fatigue driving detection.The main research contents of this paper are as follows:(1)The influencing factors of freight vehicle fatigue driving.Using the vehicle operation data under natural driving,this paper analyzes the operation characteristics of freight vehicles,including the distribution of parameters such as driving time,driving time,average speed,the number of dangerous times,and the correlation between the two parameters.And then extract vehicle operating parameters related to fatigue driving.On the one hand,it explains the severity of fatigue driving of freight vehicle drivers,and on the other hand extracts vehicle operating parameters related to fatigue driving.(2)Face detection and face feature point detection.First,the commonly used methods of face detection and face feature point detection are analyzed and summarized.Then,the histogram of oriented gradient(HOG)feature and support vector machines(SVM)algorithm are selected to complete face detection,which overcomes the problem that the face can not be accurately detected due to the illumination,partial occlusion and so on.Finally,a face feature point detection model based on the ensemble of regression trees(ERT)algorithm is established to detect eye,mouth,nose and other face feature points at millisecond level.(3)Fatigue driving detection algorithm design.A fatigue driving detection algorithm based on the fusion of vehicle operation and facial features is designed.The parameters include driving time,PERCLOS,eye blink frequency,eye closed time,and yawning times.The extraction of vehicle operating parameters is based on statistical acquisition of vehicle operating data,and the state of eyes and mouth is acquired based on the comparison of EAR and MAR values with thresholds.This algorithm realizes that when a certain parameter is not accurately detected,the fatigue state can still be accurately judged according to other parameters.