Driver Eye Scanning on Curves and on Straight Sections on Rural Highways

Abstract

Driver eye scanning behavior was investigated for two drivers on an eight mile (13km) section of a rural hilly two-lane highway. Data was collected in both directions. Nighttime data was also collected for one of the two drivers. Seven curves of various curvature and three straight level highway sections were selected for the analysis. An instrumented vehicle with a corneal reflection technique television eye scanning recording system was used. The objectives of this exploratory study were: 1) to compare the driver eye scanning behavior between left curves, right curves and straight sections, 2) to determine to what extent driver eye scanning behavior changes when a driver approaches a curve, negotiates a curve and leaves a curve, 3) to determine to what extent the radius of a curve influences driver eye scanning behavior, and 4) to what extent driver eye scanning behavior changes from daytime driving to nighttime driving. The results of this exploratory study include x-y eye fixation density maps, spatial and temporal eye scanning summary measures (including histograms) for the curve approach section, the curve section, the after curve section and the straight highway sections for day and night. Further, the x-y centers of gravity for the eye fixations are plotted for each 100 feet (30m) segment starting at 400 feet (122m) before the beginning of a given curve and ending about 400 feet (122m) after the end of a given curve. In addition, the number of eye fixations and the fixation time durations for selected objects in the driving scene (such as looking at car ahead, looking at road surface in front, looking at road environment and signs, saccades, out of views) are graphically represented as percentages in bar graphs for the various conditions and sections. Kolmogorov-Smirnov tests were used to test for statistically significant differences at the 0.05 level. The results of this exploratory study indicate that: 1) roadway geometry (left curves vs right curves vs straight sections) influences mainly the spatial eye scanning measures (eye fixation x-y centers of gravity, dispersions), 2) curves appear to be quite demanding in terms of a driver's visual information acquisition process (3.56 fixations per 100 feet or 30m at 46.8mph or 75km/h for curves vs 2.21 fixations per 100 feet or 30m at 50mph or 80km/h for straight sections), 3) the x-y centers of gravity for eye fixations indicate that a driver's eye scanning behavior starts to be influenced by a curve about 300–400 feet (91–122m) before the curve begins (drivers fixate mainly in the vicinity of the right edge line when approaching and driving through a right curve, while fixating mainly in the vicinity of the center line or left edge line when approaching and driving through a left curve in order to obtain directional and lateral position information), 4) the fixation time durations for nighttime driving are significantly longer when compared with daytime driving (0.46 seconds vs 0.39 seconds for daytime), 5) the eye fixation patterns for nighttime are more concentrated on the roadway ahead when compared with daytime, and 6) the average foveal preview distances are much shorter at night than during the day (straight road, night: AFPD=121 feet or 37m at 43.5mph or 70km/h; straight road, day: AFPD=953 feet or 290m at 50mph or 80km/h; curves, night: AFPD=131 feet or 40m at 38.6mph or 62km/h; curves, day: AFPD=577 feet or 176m at 46.8mph or 75km/h; average preview time at night for straight road = 1.90 seconds, for curves = 2.31 seconds). In conclusion the results of this exploratory study suggest that approaching and driving through a curve is a rather demanding visual task and therefore the placement of traffic signs or advertising signs within a 400 feet or 122m approach zone and the curve zone should be avoided or kept at a minimum. Further, the relatively short average foveal preview distances and the corresponding short average preview times obtained for the nighttime driving condition demonstrate that drivers operate a vehicle at night at a much lower margin of safety (in terms of reaction time and stopping distance) than during the day.