CHAPTER 3—RESULTS

To analyze the data, separate ANOVAs were conducted for each of the objects. For the pedestrians standing in turns, a between-subjects, one-way model for the four VES configurations was used. For each of the other objects, which included the pedestrians on the left and right of straight sections, the dynamic pedestrian, and the tire tread, a within-subjects, four (VES) by three (Age) mixed factorial model was used. This model was also used for the Likert scale ratings from the subjective questionnaire after the drive. In each of the models where main effects were found, Student-Newman-Keuls (SNK) tests were used to identify differences between VESs or age groups. Responses to open-ended questions were reviewed and tallied to identify emergent themes.

An =0.05 level was used to identify statistically significant effects in this report. In main effects graphs, means with the same letter are not significantly different.

OBJECTIVE MEASURES

The results for the ANOVAs conducted on the objective measures of detection distance and recognition distance for the three groups of objects—obstacle (tire tread), pedestrians in straight roadway portions, and pedestrians in curves—are presented in the following paragraphs.

Tire Tread

The analysis for the tire tread was a four (VES) by three (Age) mixed factorial design. The analyses results for detection and recognition appear in table 5 and table 6; the results indicate no significant main effects (age or VES) or interactions.

There are no statistically significant differences between the VESs for detecting or recognizing the tire tread at the =0.05 level. Figure 14 shows the mean detection and recognition distances for the tire; standard error bars are provided around the means.

Pedestrians in Straight Sections

The analysis results for the dynamic pedestrian and pedestrians standing on the right and left sides of straight roadway sections were obtained using four (VES) by three (Age) mixed factorial designs. Complete ANOVA tables for these scenarios appear in appendix D.

Age Effects

The analysis of detection of pedestrians in straight roadway sections shows significant differences in VES performance. The differences resulted from the age factor for the pedestrian on the left side of the road
(p=0.0241) but not for the pedestrian on the right (p=0.2223) or the dynamic pedestrian (p=0.0668). In particular, the SNK results show that a significant difference exists between the overall detection distance means for the younger age group (18 to 25) and the older age group (over 65). The mean detection distances across all VESs are 75.6 m (248 ft) for the younger group and 51.2 m (168 ft) for the older group. In general (for all objects), it appears that typical age effects (decreases in detection distances with increases in age) may have been present, but they were not sufficiently strong to indicate statistical differences.

VES Effects

Figure 15 illustrates the significant differences in results when pedestrians stood stationary on the left or right side of a straight section of roadway. The NIR systems resulted in significantly longer detection distances than the HLB or the FIR system. The results of the NIR 1 system show significantly longer distances for recognizing the pedestrian on the left side of the road than the results for the other three systems (NIR 2, FIR, and HLB). Both the NIR systems resulted in significantly longer distances than the results for the other two VESs for recognizing the pedestrian on the right side of the road.

Figure 16 shows the results in the dynamic pedestrian scenario. The NIR 1 system produced significantly longer detection and recognition distances than the FIR system; otherwise, the systems were undistinguished from one another in detection or recognition distances.

Pedestrians in 1,250-m Radius Curves

Each of the four pedestrian turn scenarios were analyzed using a one-way ANOVA for VES. The complete ANOVA tables for the scenarios appear in appendix E. Figure 17 presents the means and standard errors for the scenarios with a pedestrian standing on the left or right side of a left-hand turn (1,250-m (4,101-ft) radius), and figure 18 presents similar information for a right-hand turn (1,250-m (4,101-ft) radius).

Figure 17 shows that in the scenario for a pedestrian on the left side of the left-hand turn, the NIR 1 system again performs significantly better (longer distance) for detection and recognition than the HLB or the FIR. When a pedestrian was on the right side of a left-hand turn, the detection and recognition distances for each of the systems were not significantly different from each other.

Figure 18 shows that for detecting pedestrians standing on the left side of a right-hand turn (1,250-m (4,101-ft) radius), the NIR 1 system significantly outperforms the HLB and the FIR systems. Recognition distances were not statistically distinguishable between the systems. Also, in the scenario with a pedestrian on the right side of the road in a right-hand turn, the systems were not statistically distinguishable for results of either detection or recognition distance.

SUBJECTIVE MEASURES

Scaled Responses

ANOVAs were conducted on the participant responses for each of the Likert-type scale questions using the four (VES) by three (Age) mixed model described earlier. The ANOVA summary tables appear in
appendix F. A significant main effect for VES (=0.05) was found for three of the statements; significant main effects for age (=0.05) were found for two of the statements (table 13). Statement 6, which dealt with signs in the ENV IR clear weather study (ENV Volume XIII), was absent from this study because signs were not included. To maintain consistency in statement content between the present study and the ENV IR clear weather study, statements 7 and 8 are not renumbered.

Age Subjective Main Effects

In two of the statements (3 and 4), participants compared the ability of the VESs to help them stay on the road and know which direction the road was heading. It appears that the older age group gave more favorable evaluations for the VESs overall than the younger groups for these categories. In other words, older drivers tended to think that the systems tested were more advantageous than their own headlamps in helping to maintain location on the road and predicting road direction.

VES Subjective Main Effects

Results of the SNK analyses show that when asked which VES allowed them to detect objects sooner than their regular headlights (statement 1), participants indicated that the NIR 1 system was best with an average rating of 1.5. The remaining systems are above the average rating when compared to the participants’ regular headlamps, but not statistically different from each other. Similar results were found in the subjective evaluation of identifying objects (statement 2). The NIR 1 system has the best evaluation, and the others are not statistically distinguishable from each other. Figure 19 and figure 20 depict the mean responses for the subjective evaluation of detection and identification, respectively, for the four VESs. Standard error bars, as well as the SNK groupings, are shown on these graphs.

No significant differences were found when participants evaluated which VES helped them stay on the road better, know which way the road was heading, or feel safer, or which they generally thought was better than their own headlamps.

When participants were asked to evaluate the visual discomfort from the VESs compared to their regular headlamps (statement 5), the results in the model indicate a main effect for VES, but the SNK analysis does not differentiate between the VESs. Figure 21 shows the means of these responses. The mean responses were all 2.7 or less, indicating that in general the participants agreed that the VESs tested did not cause them any more discomfort than their regular headlamps.

Open-Ended Responses

Following are summaries of the comments about specific VESs made by 2 or more of the 15 participants.

FIR

• Image improvement: make clearer—can detect, but not identify (six participants).
• Image improvement: make wider (two participants).
• Hard to see in rain (lines, lanes, and/or tire) (two participants).

NIR 1

• Distracting glare (six participants).
• Like it/good system (four participants).
• Hard to see in rain (reflection) (two participants).

NIR 2

• The projector in the IR system gets in the way (three participants).
• Image improvement: make clearer—can detect, but not identify (two participants).
• Distracting glare (outweighs benefits) (two participants).
• Display should be larger (two participants).

HLB

• No comment (seven participants).
• Like it/it is good (two participants).

Previous | Table of Contents | Next

TFHRC Home | FHWA Home | Feedback United States Department of Transportation - Federal Highway Administration