U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

Report
This report is an archived publication and may contain dated technical, contact, and link information
Publication Number: FHWA-HRT-04-134
Date: December 2005

Enhanced Night Visibility Series, Volume III: Phase II—Study 1: Visual Performance During Nighttime Driving in Clear Weather

PDF Version (859 KB)

PDF files can be viewed with the Acrobat® Reader®

CHAPTER 3—RESULTS

OBJECTIVE MEASUREMENTS

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 12 (VES) by 3 (Age) by 9 (Object) mixed factorial design. ANOVA summary tables were obtained for both objective dependent measurements (table 6 and table 7). A total of 3,229 observations were obtained from the experiment for each objective measurement. When drivers were unable to detect and recognize an object, a value of 0 was assigned. Several main effects and interactions were considered significant (table 8).

Table 6. ANOVA summary table for the dependent measurement: detection distance.
Source DF SS MS F value P value  
TOTAL 3228 319608001.8      
Between
Age 2 3875582.0 1937791.0 4.29 0.0242 *
Subject/Age 27 12205792.2 452066.4    
 
Within
VES 11 8082226.5 734747.9 28.42 <0.0001 *
VES by Age 22 1277750.3 58079.6 2.25 0.0014 *
VES by Subject/Age 296 7653255.8 25855.6    
 
Object 8 180324344.1 22540543.0 679.70 <0.0001 *
Object by Age 16 1249341.1 78083.8 2.35 0.0031 *
Object by Subject/Age 216 7163085.5 33162.4    
 
VES by Object 88 9250704.6 105121.6 3.00 <0.0001 *
VES by Object by Age 176 5645156.3 32074.8 0.92 0.7750  
VES by Object by Subject/Age 2366 82880763.4 35029.9    


Table 7. ANOVA summary table for the dependent measurement: recognition distance.
Source DF SS MS F value P value  
TOTAL 3228 240107325.1      
Between
Age
2 3482588.4 1741294.2 3.97 0.0308 *
Subject/Age
27 11836662.6 438394.9    
 
Within
VES
11 5493301.2 499391 19.65 <0.0001 *
VES by Age
22 577886.9 26267.6 1.03 0.4223  
VES by Subject/Age
296 7521854 25411.7    
 
Object
8 136027251.9 17003406.5 545.65 <0.0001 *
Object by Age
16 1046045.2 65377.8 2.10 0.0094 *
Object by Subject/Age
216 6730986.1 31162    
 
VES by Object
88 5043266.7 57309.8 2.31 <0.0001 *
VES by Object by Age
176 3618771.5 20561.2 0.83 0.9481  
VES by Object by Subject/Age
2366 58728710.6 24821.9    


Table 8. Summary of significant main effects and interactions.
Source Significant Detection Significant Recognition
Between
Age
x x
Subject/Age
   
 
Within
VES
x x
VES by Age
x  
VES by Subject/Age
   
 
Object
x x
Object by Age
x x
Object by Subject/Age
   
 
VES by Object
x x
VES by Object by Age
   
VES by Object by Subject/Age
   

The main effects and most two-way interactions between age, VES, and type of object were significant (p < 0.05) for both visual-performance measurements. The VES by Age interaction lacked significance for recognition distance; this interaction was only significant for detection distance. The post hoc results for the significant main effects and interactions were graphed (figure 14 to figure 23; standard error bars are provided with the means). In the main effect graphs, means with the same letter in their grouping are not significantly different (based on the Bonferroni post hoc test).

The HLB headlamps are the most commonly available VES; therefore, the reader is urged to compare the results of other VESs to results obtained for the HLB, thus making the HLB a baseline measure. Note that this is only one halogen headlamp type and beam pattern, and not necessarily representative of all halogen lamps currently in the market.

On average, the VES by Age interaction, significant for detection distance, showed that the HOH, HHB, HLB–LP, HID, and all of the UV–A configurations with HID failed to perform better than the HLB across all three age groups (table 6). Configurations of HLB with UV–A (i.e., five UV–A, three UV–A, hybrid) across the three age groups exhibited improvements on detection distances, but these improvements averaged less than 9.1 m (30 ft); however, performance of the IR–TIS was age dependent. The younger and middle-aged drivers had farther detection distances when using the IR–TIS than when using the HLB. The younger and middle-aged participants were able to see objects 31.1 m and 45.7 m (102 ft and 150 ft) farther, respectively, with the IR–TIS compared to HLB. However, there was no improvement on detection distance for older drivers when using IR–TIS; in fact, these drivers saw, on average, 3.4 m (11 ft) farther in the HLB configuration.

Bar graph. Results on detection distances for the interaction: VES by Age. Click here for more detail.
Figure 14. Bar graph. Results on detection distances for the interaction: VES by Age.

A significant separation of results based on object contrast and the age of the driver can be observed from the Object by Age interaction for both detection and recognition (figure 15 and figure 16). For pedestrians and cyclists dressed in white clothing (i.e., high contrast), there was a significant difference between younger and older drivers’ detection and recognition. The younger drivers had the longest detection and recognition distances for all high contrast objects. The detection and recognition distances of these high contrast objects for middle-aged drivers was not statistically different from the ones for the older drivers for three out of the four objects. For pedestrians and cyclist dressed in black clothing (i.e., low contrast), there was not a significant difference between younger and middle-aged drivers, but these two age groups were significantly different from the older drivers in terms of detection and recognition distances. Both detection and recognition were shorter for older drivers than the distances for their younger counterparts. For objects that had fairly low contrast and were close to the ground (e.g., tire tread), there was no difference among age groups, and the points of detection and recognition happened close to each other.

Bar graph. Results on detection distances for the interaction: Object by Age. Click here for more detail.
Figure 15. Bar Graph. Results on detection distances for the interaction: Object by Age.


Bar graph. Results on recognition distances for the interaction: Object by Age. Click here for more detail.
Figure 16. Bar graph. Results on recognition distances for the interaction: Object by Age.

The significant difference (p < 0.05) for VES by Object under both detection and recognition distances also appears to be mainly the result of the two sets of different objects: black (low contrast) versus white (high contrast) objects (figure 17 through figure 20).

In general, the HLB performed as well as or better than the other VESs for the detection and recognition of high contrast objects (figure 17 and figure 19). The only exception was the perpendicular pedestrian with white clothing, which was detected 39.9 m (131 ft) (16 percent) farther away with the IR–TIS than with the HLB. On the other hand, the detection and recognition distances with HID were significantly different (i.e., 7 to 21 percent closer to the object) than HLB.

For low-contrast objects (figure 18 and figure 20), the HLB outperformed most of the VESs, with the IR–TIS again being the exception. The IR–TIS detected the cyclist and pedestrians with black clothing significantly farther away than the HLB; however, the other dark objects closer to ground level that do not produce heat, such as the child’s bicycle and the tire tread, were not detected or recognized as far away with the IR–TIS as they were by using the HLB.

Across all objects, the halogen baseline configuration allowed drivers to detect and recognize objects sooner than its HID counterpart. Depending on the type of object, the halogen allowed object detection ranging from 54.6 m (179 ft) (21 percent for high contrast objects) to 40.8 m (134 ft) (33 percent for low contrast objects) farther than the HIDs.

Bar graph. Results on detection distances for the interaction: VES by Object: Pedestrians and cyclists in white clothing. Click here for more detail.
Figure 17. Bar graph. Results on detection distances for the interaction: VES by Object:
Pedestrians and cyclists in white clothing.


Bar graph. Results on detection distances for the interaction: VES by Object: child's bicycle, tire tread, and pedestrians and cyclists in black clothing. Click here for more detail.
Figure 18. Bar graph. Results on detection distances for the interaction: VES by Object: child’s bicycle, tire tread, and
pedestrians and cyclists in black clothing.


Bar graph. Results on recognition distances for the interaction: VES by Object: Pedestrians and cyclists in white clothing. Click here for more detail.
Figure 19. Bar graph. Results on recognition distances for the interaction: VES by Object:
Pedestrians and cyclists in white clothing.


Bar graph. Results on recognition distances for the interaction: VES by Object: child's bicycle, tire tread, and pedestrians and cyclists in black clothing. Click here for more detail.
Figure 20. Bar graph. Results on recognition distances for the interaction: VES by Object: child’s bicycle, tire tread, and
pedestrians and cyclists in black clothing.

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 21). Both detection and recognition follow the same pattern with respect to age: distances were not significantly different between younger and middle-aged drivers or middle-aged and older drivers, but they were significantly different between younger and older drivers. The detection and recognition distances for the older drivers were the shortest; the younger drivers had the longest detection and recognition distances.

Bar graph. Bonferroni post hoc results on detection and recognition distances for the main effect: age. Click here for more detail.
Figure 21. Bar graph. Bonferroni post hoc results on detection and recognition distances for
the main effect: age.

VESs were significantly different (p < 0.05) in terms of detection and recognition distances. The post hoc analysis for the VES main effect suggests that there was a significant difference between the detection distances for the HLB baseline and the IR–TIS, where drivers with the IR–TIS were able to detect objects 24.7 m (81 ft) sooner than with HLB. Furthermore, there was a significant difference between detection distances for the HLB and the HID, HLB–LP, and hybrid and three UV–A headlamps added to the HIDs (figure 22). The HLB was able to provide for detection of objects, on average, of 21.3 to 30.2 m (70 to 99 ft) farther away than the other five VESs. There was a significant difference between recognition distances for HLB and recognition distances for any of the HID configurations and the HLB–LP. Drivers using the HLB were able to recognize objects over 18.3 m (60 ft) farther away than drivers using any of the other five configurations. Recognition distances were not significantly different between HLB and the IR–TIS or between HLB and HLB supplemented with UV–A.

Bar graph. Bonferroni post hoc results on detection and recognition distances for the main effect: VES. Click here for more detail.
Figure 22. Bar graph. Bonferroni post hoc results on detection and recognition distances for
the main effect: VES.

Post hoc tests for the type of object main effect show no significant difference among the pedestrians in white clothing and cyclist in white clothing in terms of detection. However, on average, drivers recognized the pedestrians in white clothing farther away than the cyclist in white clothing (figure 23). No significant difference was found among the pedestrians in black clothing in terms of detection or recognition. Thus, clothing contrast, rather than object motion, appears to be responsible for the significant differences observed (i.e., objects with light color clothing were detected and recognized farther away than the dark color clothing counterparts no matter if they were moving or still). While there was a significant difference between the cyclist in black clothing and the pedestrians in black clothing in terms of detection and recognition, it was probable that the increased distances for the cyclist in black clothing could be attributed to the detection of the bicycle’s rims (high specular reflectance) rather than detection of the actual cyclist given that the cyclist was wearing the same type of clothing the pedestrians were wearing; however, the cyclist in black clothing was not detected or recognized as far away as the white-clothed counterpart. The tire tread and child’s bicycle were statistically different (p < 0.05) from the other objects. The tire tread had the shortest detection and recognition distances. The detection and recognition distances for the child’s bicycle were shorter than the cyclist in black clothing but larger than the pedestrians in black clothing and the tire tread. The child’s bicycle was set on the side, centered across the right edgeline, and the rims were not facing the driver. Therefore, the drivers were not able to experience a specular reflectance similar to the one the rims of the cyclist bicycle, but the reflectance of the child’s bicycle was higher than the one for the pedestrians in black clothing and the tire tread.

Bar graph. Bonferroni post hoc results on detection and recognition distances for the main effect: object. Click here for more detail.
Figure 23. Bar graph. Bonferroni post hoc results on detection and recognition distances for
the main effect: object.

SUBJECTIVE MEASUREMENTS

An ANOVA was performed to analyze the subjective measurements taken on the Smart Road portion of the study. The model for this portion of the study was a 12 (VES) by 3 (Age) factorial design. ANOVA summary tables were generated for each of the seven subjective statements (table 9 through table 15), and the significant main effects and interactions were summarized (table 16).

Table 9. ANOVA summary table for the Likert-type rating for detection.
Statement 1: Detection
Source DF SS MS F value P value  
Between
Age
2 9.8 4.9 0.47 0.6285  
Subject/Age
27 280.0 10.4     
 
Within
VES
11 91.3 8.3 7.36 <0.0001 *
VES by Age
22 33.1 1.5 1.33 0.1472  
VES by Subject/Age
296 333.6 1.1     
TOTAL 358 747.8       


Table 10. ANOVA summary table for the Likert-type rating for recognition.
Statement 2: Recognition
Source DF SS MS F value P value  
Between
Age
2 11.1 5.5 0.53 0.5925  
Subject/Age
27 280.3 10.4     
 
Within
VES
11 54.8 5.0 4.66 <0.0001 *
VES by Age
22 28.6 1.3 1.22 0.2302  
VES by Subject/Age
296 315.9 1.1     
TOTAL 358 690.7       


Table 11. ANOVA summary table for the Likert-type rating for lane-keeping assistance.
Statement 3: Lane-keeping assistance
Source DF SS MS F value P value  
Between
Age
2 7.7 3.9 0.25 0.7802  
Subject/Age
27 415.1 15.4     
 
Within
VES
11 14.2 1.3 1.73 0.0663  
VES by Age
22 23.2 1.1 1.42 0.1040  
VES by Subject/Age
296 220.4 0.7     
TOTAL 358 680.6       


Table 12. ANOVA summary table for the Likert-type rating for roadway direction.
Statement 4: Roadway direction
Source DF SS MS F value P value  
Between
Age
2 7.0 3.5 0.24 0.7860  
Subject/Age
27 391.7 14.5     
 
Within
VES
11 10.0 0.9 1.14 0.3265  
VES by Age
22 22.9 1.0 1.31 0.1617  
VES by Subject/Age
296 235.1 0.8     
TOTAL 358 666.7       


Table 13. ANOVA summary table for the Likert-type rating for visual discomfort.
Statement 5: Visual discomfort
Source DF SS MS F value P value  
Between
Age
2 69.5 34.7 2.49 0.1021  
Subject/Age
27 377.2 14.0     
 
Within
VES
11 41.9 3.8 2.77 0.0020 *
VES by Age
22 21.0 1.0 0.69 0.8454  
VES by Subject/Age
296 407.6 1.4     
TOTAL 358 917.2       


Table 14. ANOVA summary table for the Likert-type rating for overall safety rating.
Statement 6: Overall safety rating
Source DF SS MS F value P value  
Between
Age
2 12.9 6.5 0.41 0.6687  
Subject/Age
27 427.8 15.8     
 
Within
VES
11 20.0 1.8 1.53 0.1209  
VES by Age
22 18.5 0.8 0.71 0.8338  
VES by Subject/Age
296 352.7 1.2     
TOTAL 358 832.0       


Table 15. ANOVA summary table for the Likert-type rating for overall VES evaluation.
Statement 7. Overall VES evaluation
Source DF SS MS F value P value  
Between
Age
2 12.2 6.1 0.42 0.6587  
Subject/Age
27 387.5 14.4     
 
Within
VES
11 32.3 2.9 2.02 0.0566  
VES by Age
22 26.4 1.2 0.83 0.6929  
VES by Subject/Age
296 430.5 1.5     
TOTAL 358 888.9       


Table 16. Summary of significant main effects and interactions for the Likert-type rating scales.
  Significance Summary per Statement
Source 1 2 3 4 5 6 7
Between
Age              
Subject/Age              
 
Within
VES x x     x    
VES by Age              
VES by Subject/Age              

To understand drivers’ ratings of the various VESs in terms of safety and comfort, the results for all seven statements and each VES are sorted by ascending mean rating. Drivers rated the IR–TIS as the top configuration that (1) allowed them to detect and recognize objects sooner, (2) made them feel safer, and (3) performed as the best VES. However, drivers also gave the IR–TIS the lowest (i.e., worst) rating for its effectiveness with lane-keeping assistance and also rated it as the highest producer of visual discomfort when compared to the other VESs. The HID, HHB, and HID with three UV–A headlamps were the lowest in aiding drivers to detect and recognize objects sooner, with a tendency toward a neutral rating. In addition, when ranked on the mean subjective ratings, the HLB had a higher ranking than the HID for six out of the seven statements, which suggests that it is perceived as allowing faster detection and recognition, better lane-keeping assistance, less visual discomfort, and increased safety. A list of all statements follows.

  • Statement 1: This vision enhancement system allowed me to detect objects sooner than my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    IR–TIS
    1.47
    Five UV–A + HLB
    2.47
    Hybrid UV–A + HLB
    2.50
    Hybrid UV–A + HID
    2.83
    Five UV–A + HID
    2.87
    Three UV–A + HLB
    2.97
    HOH
    2.97
    HLB–LP
    3.14
    HLB
    3.17
    Three UV–A + HID
    3.23
    HHB
    3.30
    HID
    3.40

  • Statement 2: This vision enhancement system allowed me to recognize objects sooner than my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    IR–TIS
    2.00
    Five UV–A + HLB
    2.50
    Hybrid UV–A + HLB
    2.57
    Hybrid UV–A + HID
    2.83
    Three UV–A + HLB
    2.90
    HOH
    2.97
    Five UV–A + HID
    3.00
    HLB
    3.07
    HLB–LP
    3.10
    HHB
    3.30
    HID
    3.37
    Three UV–A + HID
    3.43

  • Statement 3: : This vision enhancement system helped me to stay on the road (not go over the lines) better than my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    Hybrid UV–A + HID
    3.07
    Five UV–A + HID
    3.10
    HLB–LP
    3.14
    Five UV–A + HLB
    3.17
    Hybrid UV–A + HLB
    3.17
    HOH
    3.17
    HID
    3.17
    Three UV–A + HID
    3.30
    Three UV–A + HLB
    3.40
    HLB
    3.43
    HHB
    3.50
    IR–TIS
    3.77

  • Statement 4: This vision enhancement system allowed me to see which direction the road was heading (i.e., left, right, straight) beyond my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    Five UV–A + HLB
    3.00
    HOH
    3.07
    Hybrid UV–A + HLB
    3.17
    Five UV–A + HID
    3.20
    HHB
    3.23
    Hybrid UV–A + HID
    3.23
    HLB
    3.33
    IR–TIS
    3.40
    Three UV–A + HLB
    3.40
    HID
    3.47
    HLB–LP
    3.48
    Three UV–A + HID
    3.57

  • Statement 5: This vision enhancement system did not cause me any more visual discomfort than my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    Five UV–A + HLB
    2.20
    HLB
    2.23
    HOH
    2.23
    Hybrid UV–A + HLB
    2.30
    Three UV–A + HLB
    2.37
    Hybrid UV–A + HID
    2.40
    HLB–LP
    2.48
    Five UV–A + HID
    2.57
    HHB
    2.60
    HID
    2.77
    Three UV–A + HID
    2.90
    IR–TIS
    3.43

  • Statement 6: This vision enhancement system made me feel safer when driving on the roadway at night than my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    IR–TIS
    2.57
    Five UV–A + HLB
    2.87
    Hybrid UV–A + HLB
    2.97
    Hybrid UV–A + HID
    2.97
    HOH
    3.03
    HLB–LP
    3.10
    Three UV–A + HID
    3.13
    HLB
    3.17
    Three UV–A + HLB
    3.20
    HHB
    3.27
    Five UV–A + HID
    3.27
    HID
    3.57

  • Statement 7: This is a better vision enhancement system than my regular headlights (1=Strongly Agree; 7=Strongly Disagree).

    VES Mean Rating
    IR–TIS
    2.37
    Five UV–A + HLB
    2.57
    Hybrid UV–A + HLB
    2.73
    HOH
    2.87
    Hybrid UV–A + HID
    2.87
    HLB
    3.07
    Three UV–A + HLB
    3.13
    Three UV–A + HID
    3.13
    HHB
    3.17
    Five UV–A + HID
    3.20
    HLB–LP
    3.34
    HID
    3.40

The only significant difference for the statements was found in the VES main effect, specifically for statements 1, 2, and 5 (table 16). For statement 1—this vision enhancement system allowed me to detect objects sooner than my regular headlights-there is a significant difference (p < 0.05) between the IR–TIS configuration and all other configurations (figure 24). The IR–TIS received a mean rating of 1.42 (i.e., “Agree” to “Strongly Agree”), while other configurations remained clustered in the “Agree” range.

Post hoc results for statement 2—this vision enhancement system allowed me to recognize objects sooner than my regular headlights-again show the IR–TIS attaining the best mean rating (figure 25). The recognition rating was not as good as that given for detection, but it is still on the “Agree” range. However, while there is a significant difference (p < 0.05) in ratings between HLB and the IR–TIS, this difference does not exist between the IR–TIS and the HLB supplemented by the three UV–A configurations (five UV–A, three UV–A, hybrid). There are also no significant differences between HLB and the other 10 VESs. All the configurations remained in the “Agree” range.

Statement 5—this vision enhancement system did not cause me any more visual discomfort than my regular headlights-was also responsible for some significant differences (p < 0.05). There is a significant difference between HLB and the IR–TIS but not between HLB and the other 10 configurations. IR–TIS has a tendency toward neutral for that statement, but all other VESs align along the center of the “Agree” region (figure 26).

Bar graph. Bonferroni post hoc results on the ratings evaluating detection for the main effect: VES. Click here for more detail.
Figure 24. Bar graph. Bonferroni post hoc results on the ratings evaluating detection for
the main effect: VES.


Bar graph. Bonferroni post hoc results on the ratings evaluating recognition for the main effect: VES. Click here for more detail.
Figure 25. Bar graph. Bonferroni post hoc results on the ratings evaluating recognition for
the main effect: VES.


Bar graph. Bonferroni post hoc results on the ratings evaluating visual discomfort for the main effect: VES. Click here for more detail.
Figure 26. Bar graph. Bonferroni post hoc results on the ratings evaluating visual discomfort for
the main effect: VES.

 

Previous    Table of Contents    Next
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101