Mechanism Study On The Vision-based Speed Control And Lane Keeping

Human factor is the primary cause which causes traffic accidents. According to the Japanese statistic data, about 90% accidents were caused by the vehicle driver and the pedestrian. Vision information account for above 90% of all the information which drivers gain in the driving, drivers mainly use the peripheral vision information to percept the vehicle speed and the vehicles position, then carry on the judgment and decision-making. Obviously, to raise the road traffic safety, it is important to improve drivers' perception information to make drivers advance control vehicle speed, keep lane and real-time adopt the speed which adapts with the road geometry.Firstly, this paper analyzes the driver's visual characteristics, including: visual field, gaze as well as factors which influence visual attention. Then this paper analyzes visual factors which influence speed perception, including: the edge rate, the optical flow rate and the contrast, then analyzes the influence mechanism from temporal frequency, spatial frequency, angular speed, physiological psychology. Based on the above analysis, according to the concept of edge rate and Newton's equation of motion, the speed control design model which based on the assumption that the driver maintains the constant edge rate with the comfortable deceleration rate is deduced. By using the 3DMAX, the edge rate experiment videos that car moving on the highway are made, we design 260 stimulation of Subjectively Equal Speed experiments by using the Forced-Choice Method, through 6 males and 2 females' experiment results, we find that driver's speed perception value is the largest when the edge rate is 12HZ, but the difference between 8HZ and 16HZ is litter, considering driver's speed perception value is the largest when the edge rate is 12HZ under the fog environment, then we select 12HZ as the reasonable value. According to experiment result, we design the edge rate spacing for controlling the speed from 120km/h to 90km/h.Aiming at previous study used uniform reduction of the image contrast, we made experiment video that car moving on the highway in real fog environment with 3DMAX (the density of fog increased linearly with visual distance). 135 Subjectively Equal Speed experiments were designed by using the Forced-Choice Method, 6 males and 2 females' experiment results indicated that fog exerts a two-way influence to the driver's speed perception. When visibility was low (e.g:39m), fog resulted in overestimation of the speed, When visibility was higher, fog resulted in underestimation of the speed, it is not in accordance with one-way influence of Snowden and Anstis. Have analyzed the reason why the experimental result is inconsistent and emerge in terms of light stream rate and contrast ratio in thesis. When visibility is low, the nearby optical flow rate is great which cause the driver overestimate the speed; When visibility is higher, the decline of contrast cause the driver underestimate the speed. Finally, the improvement method which based on the edge rate was proposed, through speed perception experiment, driver's speed perception value was the largest when the edge rate was 12HZ under the fog environment, then 12HZ was selected as the reasonable value.Finally, the paper carries on the induction and summary about the existed vision-based lane-keeping theory, including: (1) lane-keeping theory which based on the optical flow and retina flow; (2) lane-keeping theory which based on the gaze point; (3) lane-keeping theory which based on the splay angle.