U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
|This report is an archived publication and may contain dated technical, contact, and link information|
Publication Number: FHWA-HRT-04-137
Date: December 2005
Enhanced Night Visibility Series, Volume VI: Phase II—Study 4: Visual Performance During Nighttime Driving in Fog
PDF Version (1.27 MB)
PDF files can be viewed with the Acrobat® Reader®
Results included in this report are based on statistically significant effects at an = 0.05 level except where otherwise stated. In main effect graphs, means with the same letter are not significantly different based on the Bonferroni post hoc test. Bars above and below the means indicate standard error.
Maintaining a consistent fog density over multiple testing sessions is difficult even with artificial fog, but doing so was important to allow unbiased comparisons between the VESs. One available metric to test fog density is the amount of backscatter, light reflected back from the fog to the driver’s eye point. Not surprisingly, different headlamp designs elicit different amounts of backscatter in the same fog condition. The consistency of fog density can be inferred by comparing these backscatter differences from multiple fog sessions to backscatter differences in a baseline clear condition; if the differences between VESs are similar in both conditions, the fog density can be considered consistent.
A baseline backscatter estimate in a no-fog, clear condition was conducted as part of the fog model development discussed in ENV Volume IX. To measure backscatter in the fog condition, an illuminance meter was positioned in the experimental vehicles to collect backscatter data during the object detection and recognition trials as discussed in the Apparatus section. This backscatter data was analyzed to determine any potential bias or potential confounding differences that existed in fog for the different VESs. Comparison of this data to the baseline estimates of each VES showed that VESs with low backscatter in the clear condition had low backscatter in fog conditions, and VESs with high backscatter in the clear condition had high backscatter in fog conditions. Therefore, when controlling for the differences in the baseline backscatter for each VES, on average, the fog provided fairly consistent levels of backscatter for each VES. As shown in table 5, the variation in fog was also similar across the VESs. This suggests that although there were variations in the fog, there was no VES that consistently had thicker or thinner fog than the other VESs. Also, the three age groups had similar backscatter ranges, with means between 0.24 lux (lx) and 0.26 lx. Given that no consistent bias in fog density could be ascertained, the variations in fog were considered random and were not analyzed further in the following objective measures. ENV Volume IX gives more information on the fog and the fog model.
An ANOVA was performed on the objective measurements taken during the Smart Road portion of the study. The model for this portion of the study was a 6 (VES) by 3 (Age) mixed factorial design. ANOVA summary tables were obtained for both objective dependent measurements (table 6 and table 7). A total of 626 observations resulted from the experiment for each objective measurement. When drivers were not able to detect and recognize a pedestrian, a value of 0 was assigned.
Both main effects were significant (p < 0.05) for detection and recognition distances, but not their interaction (table 8). The results for the main effects VES and age are shown in figure 11 and figure 12.
The HLB headlamp is the most commonly available VES, making its experimental results a baseline measure. It is important to compare the results of other VESs to results obtained for the HLB in the following descriptions of the significant results. Note that this is only one halogen headlamp type and beam pattern; it does not necessarily represent all halogen headlamps currently on the market.
VESs were significantly different from each other (p < 0.05) in terms of the detection and recognition distances (figure 11). Post hoc analyses showed that HLB provided detection and recognition distances that were significantly longer than the HID VES by approximately 8.5 m (28 ft). HLB was not statistically different from the other four VESs in terms of detection or recognition distances.
Figure 11. Bar graph. Bonferroni post hoc results for the main effect: VES.
The results from the ANOVA showed a significant difference (p < 0.05) among the three age groups in terms of detection and recognition distances (figure 12). Both detection and recognition follow the same pattern with respect to age: distances were not significantly different between older and middle-aged drivers, but they were significantly different between these two age groups and the younger drivers. The detection and recognition distances for the two older age groups were longer than those of the younger drivers by approximately 8.5 m (28 ft).
Figure 12. Bar graph. Bonferroni post hoc results for the main effect: age.
An analysis of variance (ANOVA) was performed to analyze the subjective measurements taken on the Smart Road. The model for this portion of the study is a 6 (VES) by 3 (Age) factorial design. Table 9 through table 15 present the ANOVA summaries for each of the seven subjective statements, and table 16 summarizes significant main effects and interactions.
Table 9. ANOVA summary table for the Likert-type rating for detection.
Table 10. ANOVA summary table for the Likert-type rating for recognition.
Table 11. ANOVA summary table for the Likert-type rating for lane-keeping assistance.
Table 12. ANOVA summary table for the Likert-type rating for roadway direction.
Table 13. ANOVA summary table for the Likert-type rating for visual discomfort.
Table 14. ANOVA summary table for the Likert-type rating for overall safety rating.
Table 15. ANOVA summary table for the Likert-type rating for overall VES evaluation.
To understand drivers’ ratings of the various VESs in terms of safety and comfort, the results of all seven statements for every VES were sorted by ascending mean rating. Drivers rated the IR–TIS as the most likely to help them detect and recognize pedestrians sooner. In general, HIDs received better rankings than did HLBs on statements relating to farther detection, effectiveness in lane-keeping assistance, less visual discomfort, and overall perception of safety. A list of all statements and mean ratings for each VES is presented next.
Statistically significant post hoc test results were graphed for ease of interpretation (figure 13 through figure 15). There were no significant differences in terms of age. VES type was significant (p < 0.05) for statements 1, 2, and 7. In statement 1, “This vision enhancement system allowed me to detect objects sooner than my regular headlights,” a significant difference (p < 0.05) was observed between the IR–TIS configuration and all other configurations. IR–TIS received a mean rating of 1.33 (i.e., “Strongly Agree”), while the HLB baseline received a mean rating of 2.50 (figure 13). HLB and HID were not statistically different from the other headlamp configurations or from each other. Statements 2 and 7 show that HLB and HID were not statistically different from the other headlamp configurations or from each other (figure 14 and
Figure 13. Bar graph. Bonferroni post hoc results on the ratings evaluating detection for the main effect: VES.
Figure 14. Bar graph. Bonferroni post hoc results on the ratings evaluating recognition for the main effect: VES.
Figure 15. Bar graph. Bonferroni post hoc results on the overall rating for the main effect: VES.