A dynamic driving simulation investigated the visibility of realistic targets with variable levels of artificial illumination. 20 video clips were recorded on the Virginia Smart Road using a high resolution digital camera. Test clips contained roadway targets such as deer, tires, and pedestrians who entered the road from either the left or right sides. Road illumination levels varied from clip to clip: some included overhead luminaires, others were illuminated only by the vehicle's low-beam headlights. 16 licensed drivers watched the video clips and responded when they recognized a target. Participants response times were recorded, and their verbal identifications of target type and location were documented by the experimenter.The results showed that target size and motion significantly affected recognition time in dark conditions (no luminaires): walking pedestrians wererecognized at the greatest distance followed by stationary pedestrians, deer, and the tire. Recognition was also affected by the side of the road from which pedestrians entered: pedestrians entering from the right were recognized from significantly longer distances than those entering from theleft. Furthermore, recognition times for pedestrians under overhead lighting were significantly longer than for pedestrians illuminated only by headlights. These findings provide a preliminary assessment of the effects of target structure, motion, and lighting on drivers recognition of realistic objects. These results will be compared with performance of drivers currently being tested while driving on the Smart Road to clarify the validity of video-based night driving simulators.
Abstract