Research On Driver Visual Perception Character And Its Modelling

The driver is the core of the driver-vehicle-road closed-loop system and undertakes multiple tasks,such as environment observation,decision making and vehicle handling.Among them,the full observation of the road and traffic environment is the premise and basis for correct decision-making and manipulation;and vision is the main way for drivers to obtain information,about 85% of the information is obtained visually.Therefore,the driver’s visual perception characteristics are directly related to the operation state of the driver-vehicle-road closed-loop system.Accurate and sufficient visual perception is the key to the stable operation of the closed-loop system.Therefore,this paper aims to study driver visual perception characteristics during driving.The main research work and results are as follows:(1)Given the consideration of the difficulty in obtaining road alignment in field test,and the difficulty to reflect driver’s visual perception characteristics in the driving simulator test due to its simple traffic environment,field and driving simulator tests were designed in this study.The field test was used to explore and discover driver visual characteristics in practical traffic environment,and it was also used as an effective verification of driving simulator test.The driving simulator test was used to collect driver’s eye movements,vehicle operating conditions,and road and environmental information.The collected test data was divided into several parts according to road alignment(straight line,left turn and right turn),and the traffic environment around the vehicle is marked for each road section.The raw eye movement data was processed by frequency correction,interpolation,filtering,etc..In the end,the real-time driver gaze point is obtained by the I-VT method.(2)Driver visual characteristics on straight line and curved(left and right)sections were studied.The front scene collected by the eye tracker was divided into several AOIs(area of interests).The distribution of the driver’s gaze point and gaze time in AOIs was analyzed.The distribution probability and probability density of gaze point in the horizontal direction were investigated.Combined with the real-time distribution position of the fixation point on the foreground image,driver visual feature points on straight line and curved sections were obtained,and the distribution of driver attention to the visual feature point was further analyzed.According to the difference on visual feature point betweeen straight line and curved sections,the phenomenon of “gaze jump” at the intersection of straight line and curved sections was proposed.The transition section of the visual characteristics was defined,and the visual perception feature on the transition section is analyzed.Using time series cross-correlation analysis,the cross-correlation between driver gaze yaw angle and vehicle steering wheel angle was analyzed.It was concluded that the change of the steering wheel angle conincides with the change of driver’gaze yaw angle,but it lags behind.This phononmen reveals the guiding effect of driver visual character on vehicle handling.(3)In order to obtain the real-time driver visual feature point,the driving simulator test data was used to solve the real-time driver preview time on straight line and curved sections.The influence of driver attribute,road curvature and driving speed on driver preview time were analyzed.It turned out that raod curvature is a key factor affecting t driver preview time.Considering the fan-shaped divergence distribution of preview time along road curvature,statistical method was used to estabilish the relation between driver preview time and road curvature.The normality test,data transformation and the resampling processing were performed on the raw preview time data.Using the processed data,the regression model of preview time with respect road curvature was constructed.The estimation accuracy and prediction accuracy of the regression model were also analyzed.In the case that the sample size is not rich enough,the raw preview time data were processed by the method of group median,and the exponential model of preview time in terms of road curvature was constructed.The differences between the regression model and the fitted model were compared and analyzed,and the main reasons for the differences were discussed.(4)The influence of driver visual perception strategy(preview strategy)on vehicle track following was investigated by the feedforward optimal curvature model and the feedforward and feedback integrated predictive control model.While using the optimal curvature driver model,the strategy of preview time changes with road curvature can significantly improve vehile track following performane more than 50% compared with the strategy of fixed preview time.While using the model predictive driver model,the difference in vehicle track following performance under the two different preview strategies was quite small,but the strategy of preview time changes with road curvature can significantly improve the calculation speed(about 25 %),which will reduce driver decision load.The research reveals the guiding mechanism of driver visual perception characteristics on vehicle handling.The obtainment method of real-time driver preview time was proposed,and driver visual feature point was obtained.Besides,the role of driver visual perception strategy on vehicle handling was analyzed.