Multi-scene Driving Stability Judgment Based On Driver Status Monitoring

Since the invention of the automobile,safe driving has always been a key research project in the automotive field.While exploring the field of unmanned driving,intelligent assisted driving technology has begun to make important contributions to ensuring driving safety.Traditional fatigue driving detection only pays attention to the driver’s mental state and ignores the huge differences in different driving environments.Bad driving environment often requires the driver to have a more sensitive response and control ability,comprehensive consideration of the driving environment to determine driving stability is an important means to improve driving safety.In this paper,machine vision is used to detect the driver’s mental state,visibility,and the slippery degree of the road,and the three types of data measured are fused at the decision-making level to determine driving stability.For different driving conditions,appropriate measures should be taken to reduce the accident rate,the main contents are as follows.(1)In order to detect the mental state of the driver,this paper designs a pair of glasses for collecting eyelid information.The PERCLOS,a kind of fatigue parameter is extracted from the driver’s eyelid information,and determines the driver’s fatigue degree by analyzing the PERCLOS value.To this end,this paper uses the convolutional neural network and the ERT regression tree algorithm to train model to detect the eyelid key point.The ERT regression tree has a better performance in the task which detect key points of the eyelid,and the accuracy reaches 92%.(2)Starting from the atmospheric scattering model,calculate the atmospheric transmittance in the image through the dark channel a priori theory,and calculate the visibility based on the distance of the observation object.A distance measurement method based on the width of the straight lane line is proposed,and the estimated visibility is consistent with the local meteorological visibility level.(3)According to the sliding friction coefficient between tires and the road surface,this paper divides the degree of road wetness into four grades,corresponding to the four types of road conditions: dry road,water-filled road,muddy road,and snow-covered road.The SVM model is constructed to classify the features of the extracted wet and slippery images.The model’s recognition accuracy of the four road conditions has reached more than 90%.(4)Finally,through the D-S evidence theory,the driver’s mental state,visibility,and the slippery degree of the road are integrated at the decision-making level to determine driving stability,and take corresponding measures based on different stability states to achieve the purpose of reducing the incidence of accidents.The simulation driving experiment proves the validity of the driving stability judgment model.