Evaluation of The Impact of Spectral Power Distribution on Driver Performance

CHAPTER 6. Overhead-Lighting Level Experiment

Introduction

The goal of the overhead-lighting level experiment was to investigate the interaction of overhead-lighting level and vehicle headlamps. The performance of a lighting system is based on the adaptation level of the driver, so it is critical to identify whether the dominant component of a driver’s visual field is the vehicle headlamps or the overhead-lighting system. This experiment was designed to investigate the relative dominance of overhead lighting and headlamps with regard to visibility.

Only the 6,000-K LED lighting was used in this experiment because, of the three lighting types used in this project, its spectrum was the most efficient in the mesopic region. Consequently, it provided the best visibility at the lowest illuminance. Eleven lighting levels were tested, ranging from off to the 100 percent.

Pavement type, color, and reflectance all affect object contrast and visibility. Therefore, experiments were performed on two types of pavement: lighter-colored concrete and darker asphalt. The position of the target with respect to the overhead-lighting masts also affects visibility. By combining target placement with respect to the luminaire and pavement type, two worst-case scenarios for target visibility were created to test minimum VLs. In the first case, a target was placed on dark asphalt in a position with low VI. In the second case, a target was placed on light concrete in a positon with high VI. Pedestrians were also used to test visibility, representing a real-life dangerous driving scenario.

To get a detailed picture regarding visibility at the various lighting levels, both targets and pedestrians were used to measure detection distance. Only gray clothing and targets were used, because this experiment focused on overhead-lighting level and mesopic vision, not object color. Participants detected the targets and pedestrians. To add a level of detail, participants were also asked to recognize which direction the targets and pedestrians were facing.

Tests were performed with headlamps on to simulate real-life driving scenarios and with headlamps off to examine the effect of overhead lighting alone.

Research Objectives

The research objectives of the overhead-lighting level experiment were to evaluate the following:

• Impact of headlamps driver visual performance.
• Impact of overhead-lighting level on driver visual performance.
• How headlamps and overhead lighting interact to affect driver visual performance.
• Minimum overhead-lighting level where driver visual performance is not compromised.

In addition to the stated research objectives, the results of this experiment informed the remaining experiments regarding adaptation luminance in mesopic conditions. This information was the basis for light levels and object locations used in subsequent experiments in this project.

Experimental Design

An experiment was designed to measure the relationship of overhead-lighting level and headlamps on nighttime driving visual performance in mesopic conditions. Variables used in the experiment are listed in table 20 and table 21.

Table 20. Overhead-lighting level experiment independent variables.

Table 21. Overhead-lighting level experiment dependent variables.

All variable combinations were tested except for the headlamp-off and the overhead-lighting-off conditions together. In that combination, no lighting would be present, which would present a significant safety hazard.

Independent Variables

Age

Participants were divided into the same age groups as in previous experiments: drivers (25–35 years old) and older drivers (65 years old and older). These ages were selected on the basis of visual ability and driving experience. Younger drivers typically have better vision for driving at night, while the older population has more experience with driving in general.

Vehicle Headlamps

Vehicle headlamps were either turned on or off; when on, they were used with the neutral-density low-intensity filters described in chapter 3.

Overhead-Lighting Level

The overhead roadway lighting used was the 6,000-K LED system. Dim levels varied in 10‑percent intervals from 0-percent illuminance to 100-percent illuminance. The horizontal illuminances were measured on the roadway surface at the target II location. (Target positioning will be described in the next section.) The 100-percent level was 10 lx (0.93 fc). The 10-percent level was actually 1.37 lx (0.13 fc) because the control system and luminaire characteristics do not allow for lower dimming levels. The headlamps used were the white/yellow filters with the 30-percent neutral density filters. Vehicle headlamps were either turned on or off.

Target Type

This experiment used gray-clad pedestrians and gray targets. The pedestrians were positioned on the shoulder, 0.03 m (1 ft) away from the road, on the same side of the road as the overhead-lighting system. The targets were placed on the shoulder, also 0.03 m (1 ft) away from the road but on the opposite side of the road from the overhead lighting.

The target placement longitudinally along the roadway was selected to produce worst-case scenarios in terms of target contrast—one with high VI and one with low VI. The pedestrians were located so that the VI at head level matched that of the high VI targets. Target and pedestrian placement is described in detail in the following paragraphs.

Target I Placement:

The placement of target I attempted to create a low VI condition. It was placed on a dark asphalt surface and positioned directly across from a luminaire (figure 76), so very little light from the luminaire shone directly on the face visible to the oncoming vehicle.

Figure 76. Diagram. Overhead-lighting level experiment—target I (low VI) placement.

Target II Placement:

The placement of target II attempted to create a high VI condition. It was placed on a light asphalt surface halfway between two luminaires (figure 77), the area of maximum VI with respect to the luminaire, and as much light as possible shone on the face visible to the oncoming vehicle.

Figure 77. Diagram. Overhead-lighting level experiment—target II (high VI) placement.

Pedestrian Placement:

The placement of the pedestrian with respect to the luminaires was chosen so that the VI at approximately face height, about 1.5 to 1.8 m (5 to 6ft), matched the VI at target II, the high-VI target.

To do so, first the VIs at target II’s position were measured. Measurements were repeated with the overhead lighting varying in 10-percent intervals from 0 to 100 percent. Then, VI measurements were taken at various locations on the same side of the road as the luminaires and at 1.5 to 1.8 m (5 to 6 ft) (head height). The pedestrian was positioned where these VI measurements were approximately equal to those at target II for the same illumination condition. Table 22 lists illuminance measurements at the target I, target II, and the best-match pedestrian position. Figure 78 graphs those illuminances.

Table 22. Overhead-lighting level experiment—VI of targets and pedestrian with luminaires and no headlamps.

Figure 78. Graph. Overhead-lighting level experiment—VI on objects by overhead-lighting level.

Dependent Variables

Detection and orientation-recognition distances were measured in this experiment. Color-recognition distance was not measured; all objects were gray.

Object Luminance and Contrast

Photometric images of the pedestrians and targets were taken post hoc with a calibrated ProMetric® Radiant Imaging® camera mounted inside the test vehicle. Images were taken with headlamps on and off. For the targets, photometric images were recorded every 15 m (50 ft) out to 122 m (400 ft) from the target. For pedestrians, images were recorded every 30 m (100 ft) out to 244 m (800 ft) from the pedestrian.

The images taken were analyzed using Radiant Imaging® software. A data reductionist traced the contour of the pedestrian or target, and the software calculated the average luminance within that polygon. The luminances of the targets, pedestrians, and their backgrounds were measured to calculate contrast.

Methods

Participants

Twenty-four participants took part in the study. However, data for two of the participants were unusable because an anomaly during data collection rendered that data unreliable. Of the 22participants whose data were kept, 12 were in the older group, and 10 were in the younger group. There were 11 males and 11 females. Mean and standard deviation of participant age, visual acuity, mesopic visual acuity, and low contrast visual acuity are listed in table 23.

Table 23. Overhead-lighting level experiment participant characteristics.

Procedure

Data Collection

Upon arrival for an experimental session, participants were directed to the test vehicles, familiarized with their operation and the experiment, and then asked to proceed to the Smart Road for the experiment. As in previous experiments, two participants in two vehicles completed the experiment at one time. Participants drove 22 laps on the Smart Road, including one practice lap. One lap was from turn 2 to turn 3, as shown in figure 23. In the remaining 21 laps, the same objects (targets and pedestrians) were in the same location for each lap. Participants were told the objects would be in the same position for each lap and were reminded to say when they could see the objects and object orientations, not when they expected to see the objects. On-road researchers changed the overhead-lighting configuration and object orientations, and in-vehicle researchers turned the headlamps on and off and recorded detection and orientation-recognition. There were no laps without headlamps and without overhead lighting to avoid creating hazardous test conditions.

Data Analysis

After video reduction of the data, an ANOVA was conducted for analyses related to detection and orientation-recognition distances. An ANCOVA was conducted for analyses related to contrast.

The photometric results were also analyzed in terms of contrast and average detection distance for the objects.

Results

Lighting Characteristics

The distance at which headlamps contribute illuminance to a small target was determined by measuring VI using a light meter placed in front of the target at ground level with the vehicle headlamps on. Luminaire dim levels did not noticeably affect the VI at the target while headlamps were on. When the vehicle was within 91.4 m (300 ft) of the target, there was only about a 10-lx (0.93 fc) difference in VI between no overhead lighting and 100-percent overhead lighting at each distance. VI measurements by distance from target are graphed in figure 79.

1 m = 3.3 ft

Figure 79. Graph. Overhead-lighting level experiment—VI at target, vehicle with headlamps on at between 91 and 8 m (300 and 25 ft) from target, luminaires from 0 to 100percent.

Detection Distance Versus Lighting Conditions

The data were analyzed using a mixed-models analysis to determine significant effects of headlamp lighting condition and overhead-lighting level on detection and orientation-recognition distances.

Significant Results Summary

Separate analyses were run for the targets and the pedestrian to isolate the effects of lighting. Overhead-lighting level significantly affected pedestrian-detection distances. Overhead-lighting level significantly affected detection distance for both targets, as did headlamp condition, but there was no effect from interaction between the two (table 24).

Table 24. Overhead-lighting level experiment—significant effects on detection distance of targets and pedestrian.

Detection Distance Versus Overhead-Lighting Level

An SNK test was conducted to isolate the extent that overhead-lighting level affected detection distance, with headlamps off and with headlamps on, for the targets and pedestrians. In the figures reporting results of SNK tests, bars sharing a letter do not significantly differ from each other.

Pedestrian, Headlamps Off:

For pedestrian detection with headlamps off, 100-percent overhead-lighting level provided the greatest mean detection distance. However, although lighting levels of 50through 90 percent were not significantly different from 100 percent, they had mean detection distances significantly greater than for the 10- to 30-percent overhead-lighting levels. The same was true for the dimmest levels; 10 percent had the shortest mean detection distance, but this was not significantly different from 20- and 30-percent overhead lighting. The general trend was for increasingly intense overhead lighting to correspond with greater mean detection distances, except for 30 percent, which had an unexpectedly short mean detection distance given the general trend. Those results are shown in figure 80.

1 m = 3.3 ft

Figure 80. Chart. Overhead-lighting level experiment—SNK groupings for pedestrian-detection distance with headlamps off by overhead-lighting level.

Pedestrian, Headlamps On:

For pedestrian detection with headlamps on, the SNK results show a sharp decrease in mean detection distance when the overhead-lighting level drops from 30 to 20percent (figure 81).

1 m = 3.3 ft

Figure 81. Chart. Overhead-lighting level experiment—SNK groupings for pedestrian-detection distance with headlamps on by overhead-lighting level.

Target I, Low VI, Headlamps Off:

There were no significant differences in target I mean detection distances for the different overhead-lighting levels, but a trend of greater mean detection distances with greater overhead lighting is apparent (figure 82).

1 m = 3.3 ft

Figure 82. Chart. Overhead-lighting level experiment—SNK groupings for target I detection distance with headlamps off by overhead-lighting level.

Target I, Low VI, Headlamps On:

The target I mean detection distances were significantly different for the different overhead-lighting levels when headlamps were on. There was a significant difference in mean detection distances between 100 and 10 percent and between 0and levels 80 percent and higher, as illustrated in figure 83.

1 m = 3.3 ft

Figure 83. Chart. Overhead-lighting level experiment—SNK groupings for target I detection distance with headlamps on by overhead-lighting level.

Target II, High VI, Headlamps Off:

The mean detection distances for target II with headlamps off did not show a strong trend of greater detection distance with more intense overhead lighting. The 80-percent overhead-lighting level had a significantly greater mean detection distance than the 30-percent level, as shown in figure 84, but neither differed significantly from any of the other lighting levels.

1 m = 3.3 ft

Figure 84. Chart. Overhead-lighting level experiment—SNK groupings for target II detection distance with headlamps off by overhead-lighting level.

Target II, High VI, Headlamps On:

The mean detection distances for target II with no headlamps are similar to those with headlamps; the 80-percent level had a significantly greater mean detection distance than the 30-percent level, but neither differed significantly from any of the other overhead-lighting levels (figure 85).

1 m = 3.3 ft

Figure 85. Chart. Overhead-lighting level experiment—SNK groupings for target II detection distance with headlamps on by overhead-lighting level.

Detection Distances Versus Headlamps On or Off

Mean detection distances were greater when the vehicle’s headlamps were off for all objects—target I, target II, and the pedestrian—although the difference was not significant for the pedestrian. Those differences are shown in figure 86.

1 m = 3.3 ft

Figure 86. Chart. Overhead-lighting level experiment—mean detection distance for all objects by headlamp on/off.

Detection Distance Versus Participant Age

Young subjects detected the pedestrians and both targets at significantly greater mean distances than older participants, when all overhead-lighting levels and headlamp condition were combined. These results, illustrated in figure 87, do not differentiate between the different lighting levels.

1 m = 3.3 ft

Figure 87. Chart. Overhead-lighting level experiment—mean detection distance for all objects by participant age and lighting conditions combined.

Comparison Among Pedestrian, Target I, and Target II Detection Distances

Most pedestrian-detection distances were from more than 122 m (400 ft), but the test vehicle headlamps only contributed to object illumination at between 76.2 and 107 m (250 and 350 ft) away, so overhead lighting, more than headlamps, drove pedestrian visibility. Pedestrian-detection distances were also significantly greater with headlamps off. In addition, detection distances for the pedestrians were not significantly different from each other for overhead-lighting intensities between 60 and 100 percent, as shown in figure 81, figure 82, and figure 88. This shows there might be an opportunity to conserve energy without reducing pedestrian safety.

The target I (low VI) mean detection distances tended to be greater than those for target II (high VI), with headlamps both on and off, because target I had greater contrast with the background than did target II (figure 89). Target detection distances tended to increase with increasing overhead-lighting levels in both headlamp conditions, but the increase was more erratic with target II because of its low contrast and poor general visibility to the participants. Targets were consistently detected from farther away with headlamps off, likely because the headlamps washed out the background behind the target, reducing contrast.

1 m = 3.3 ft

Figure 88. Chart. Overhead-lighting level experiment—mean detection distances for pedestrians by overhead-lighting level.

1 m = 3.3 ft

Figure 89. Chart. Overhead-lighting level experiment—mean detection distances for targets by overhead-lighting level.

Orientation-Recognition Distance

Object orientation was randomized, unlike detection distance, because the targets were kept at the same place on the roadway. Because it was randomized, orientation-recognition distance is less likely to have participant-created error.

Pedestrian

Orientation-recognition distances varied more widely with overhead-lighting level with headlamps off than with headlamps on, because without headlamps, overhead-lighting level alone drove visibility (figure 90). At overhead-lighting levels between 0 and 30 percent and no headlamps, the pedestrians were less visible, resulting in shorter detection distances. At those same overhead-lighting levels, however, and with headlamps on, pedestrian orientation was recognized from farther away—about 91.4 m (300 ft)—the limit of the headlamp’s range. Therefore, headlamps create a boundary at about 91.4 m (300 ft) where objects will most likely be visible. Adding overhead lighting increased orientation-recognition distances beyond that 300ft (91.4 m).

Orientation-recognition distances for the pedestrian varied widely, and that variance was even wider at higher overhead-lighting levels, whether or not the headlamps were on.

1 m = 3.3 ft

Figure 90. Chart. Overhead-lighting level experiment—mean recognition distance of pedestrian by headlamps on and off and overhead-lighting level.

Target I

Figure 91 shows that for target I, the span of orientation-recognition distances with headlamps off and 10-percent overhead lighting was from 15 to 125 m (50 to 500 ft). With headlamps on, that span was from 24 to 61 m (80 to 200 ft).

With headlamps off and high overhead-lighting levels, participants recognized target orientation from distances beyond where headlamps normally illuminate. With headlamps on and overhead-lighting levels at 30 percent and below, participants recognized targets from farther away. Therefore, headlamps helped drivers see targets at those lower lighting levels.

1 m = 3.3 ft

Figure 91. Chart. Overhead-lighting level experiment—mean recognition distance of targetI by headlamps on and off and overhead-lighting level.

Target II

The orientation-recognition distances for target II vary widely with headlamps on and off at between 100- and 0-percent overhead lighting (figure 92). One trend is that with headlamps on and with between 20- and 50-percent overhead lighting, participants recognized target II orientation from farther away than with headlamps off.

1 m = 3.3 ft

Figure 92. Chart. Overhead-lighting level experiment—mean recognition distance of targetII by headlamps on and off and overhead-lighting level.

Luminance

The luminance and contrast measures were measured to determine whether detection distance correlated with changes in contrast. If so, photometric measurements could be used to predict driver visual behavior.

Pedestrian

Pedestrian-detection distances differed significantly with the 100-, 50-, and 1- percent overhead-lighting levels, so those lighting levels were chosen for the luminance analysis. Luminance for those overhead-lighting levels was measured from 30.4 to 244 m (100 to 800 ft) in 15.2-m (100‑ft) intervals. Results are shown in figure 93 and described following the figure. Log luminance is used to better show the differences in the conditions without headlamps.

1 cd/m² = 0.3 fL
1 m = 3.3 ft

Figure 93. Graph. Overhead-lighting level experiment—pedestrian log luminance by overhead-lighting level, headlamp condition, and distance.

Overhead Lighting at 100 Percent:

The luminance between 213 and 244 m (700 and 800 ft) was 0.096 cd/m² (0.028 fL) higher with headlamps on, showing some headlamp illumination was reaching the pedestrian even at 244 m (800 ft), probably via reflection off the guard rails and roadway surface. Headlamps illuminated the pedestrian more directly beginning at 122 m (500 ft), and then luminance increased sharply at 91.4 m (300 ft) from the pedestrian, where the headlamps began to directly illuminate them.

Overhead Lighting at 50 Percent:

Dimming the overhead lighting from 100 to 50 percent affected the pedestrian luminance; at 244m (800 ft) and without headlamps, the pedestrian luminance at 50-percent overhead lighting (0.145 cd/m² (0.042 fL)) was half that at 100-percent overhead lighting (0.288 cd/m² (0.084 fL)). With headlamps, the luminance at 50-percent overhead lighting was 58percent (0.226 cd/m² (0.066 fL)) of that at 100-percent overhead lighting (0.384 cd/m² (0.112 fL)), showing the percent contribution of headlamp light to luminance was different for the various overhead lighting intensities. This could be because at 100-percent overhead lighting level, the overhead lighting drowned out the headlamps’ effect. Also, slightly different vehicle and pedestrian positions between measurements could affect luminance measurements.

Overhead Lighting at 10 Percent:

Dimming the overhead lighting from 100 to 10 percent affected the pedestrian luminance; at 244m (800 ft) and without headlamps, the pedestrian luminance at 10-percent overhead lighting (0.045 cd/m² (0.013 fL)) was 15 percent of that at 100-percent overhead lighting (0.288 cd/m² (0.084 fL)) with the headlamps off. With headlamps, the luminance at 10‑percent overhead lighting was 35 percent (0.1375 cd/m² (0.040 fL)) of that at 100-percent overhead lighting (0.384cd/m² (0.112 fL)). This shows that headlamp illumination affected pedestrian luminance at that distance.

Another important aspect of this comparison is that the headlamp seemed to dominate the target light level at 152 m (499 ft), where the target luminances from each of the dim levels began to converge.

Target I (Low VI)

The 40-percent overhead-lighting level had considerably shorter detection distances than more intense levels and appears to be the level where overhead lighting and headlamp light contribute similarly to luminance. Therefore, the 100- and 40-percent levels were selected for luminance analyses. Luminance measurements were taken from 15.2 to 122 m (50 to 400 ft) in 15.2-m (50‑ft) intervals, and the results are shown in figure 94.

1 cd/m² = 0.3 fL
1 m = 3.3 ft

Figure 94. Graph. Overhead-lighting level experiment—target I luminance by overhead-lighting level, headlamp condition, and distance.

Results for 100- and 40-percent overhead lighting were similar, and headlamp condition and vehicle distance to target both affected target I’s luminance. For headlamps on and off, the targets had nearly equal luminance between 91.4 and 122 m (300 and 400 ft). At 76.2 m (250 ft) and closer, the luminances diverged, with target I in the headlamp-on condition brighter than in the headlamp-off condition. Luminances were slightly greater with headlamps on at 100-percent overhead lighting than those with headlamps on at 40-percent overhead lighting at distances closer than 60m (200 ft).

Target II (High VI)

Like target I, the luminance and contrast of target II were analyzed at 100- and 40-percent overhead-lighting levels. Results are illustrated in figure 95.

1 cd/m² = 0.3 fL
1 m = 3.3 ft

Figure 95. Graph. Overhead-lighting level experiment—target II luminance by overhead-lighting level, headlamp condition, and distance.

Results for 100- and 40-percent overhead lighting were similar, and headlamp condition and vehicle distance to target both affected the target II’s luminance. In both overhead-lighting conditions, target luminance with headlamps on was greater than with headlamps off, even at longer distances. Like target I, luminance increased rapidly as measurements were taken from 60m (200 ft) and closer with the headlamps on because headlamp light contributed to luminance. At greater distances, headlamp contribution to luminance was smaller for target II, but the headlamps-on and headlamps-off curves never converged, showing that even at great distances, headlamps contributed slightly to target luminance.

Discussion

Although this experiment did not specifically evaluate models of mesopic vision—a project objective—its results provide insight on human visual behavior while driving in mesopic conditions as a method to determine the adaptation level of the driver, which is an input to the mesopic model. Pedestrians and small roadside targets were largely visible to participants even in low contrast conditions and without headlamps. Dimming the roadway lighting did not affect luminance ratio and contrast of objects, but object visibility, as measured by detection and object-recognition distances, did change depending on roadway lighting dim level. At 40-percent overhead lighting and below, detection distances were much shorter than at higher lighting levels. At 30-percent overhead lighting and below, orientation-recognition distances were much shorter. The range of orientation-recognition distances narrowed considerably at 30-percent overhead lighting and below. Therefore, participants were able to detect and recognize low-contrast objects, but that ability dropped off at reduced overhead lighting.

Viewing angle affected object visibility, as did the angle of light on the object. Object illuminance combined with contrast affected visibility in terms of quantity, in accordance with Adrian’s model.⁽⁵⁴⁾

Contrast polarity also affected object visibility. The negatively contrasting target—target I—had greater detection distances than the positively contrasting target—target II. Adrian’s model accounts for the tendency for negatively contrasting objects to have higher visibility.⁽⁵⁴⁾ However, when the vehicle approached the negatively contrasted target, the headlamps illuminated it, increasing its luminance with respect to the background. As the vehicle approached, the target contrast was close to zero for a period of time, rendering it invisible. When the vehicle was close enough, the headlamps illuminated the target enough to produce positive contrast. Thus, negatively contrasted objects are initially more visible than positively contrasted objects, but as a vehicle with headlamps approaches them, there is a span of distance where the object has low contrast and very poor visibility. Positively contrasted objects only increase in contrast as a vehicle approaches.

Zones of Visibility

Headlamps and roadway lighting combine to form three zones of visibility for object detection. These zones, introduced by Boyce, include near, intermediate, and far categories.⁽⁹⁵⁾ The near zone is where visibility is dominated by headlamp light. The intermediate zone is where headlamp and overhead lighting combine to affect visibility. The far zone is where overhead lighting alone contributes to visibility.

The zones of visibility identified in this experiment for pedestrians and targets are listed in table 25. The near zone for target visibility was aligned with previous research that found headlamps were the dominant light source up to approximately 40 m (130 ft).⁽⁹⁵⁾ The differences in visibility zones between the pedestrian and the targets are because they are different sizes and are different distances from the ground. The boundaries for the intermediate zone for pedestrians were not as clear from the data because the pedestrian contrast polarity shift was more gradual than that of the targets.

Table 25. Overhead-lighting level experiment—zones of visibility for pedestrian and targets.

Stopping Sight Distance

Lighting design should consider stopping sight distance, because illuminated objects must be visible from far enough away for drivers to stop. The detection distances were assumed to be stopping sight distances, and vehicle speeds corresponding to those distances were calculated using the stopping sight distance formula. The following calculations take into account both headlamp and overhead light, meeting a project objective of evaluating influence of the interaction of vehicle headlamps and overhead lighting on driver visual performance. The results are listed in table 26.

Table 26. Overhead-lighting level experiment—object stopping sight distance and corresponding speed.

There are many instances in which the minimum detection distance was shorter than suggested by stopping sight distance at a typical speed limit of about 24 km/h (15 mi/h). The implication is that, for those situations, drivers typically cannot see objects in time to stop. That was the case for many conditions using the targets but not so with the pedestrian, where drivers could travel as fast as 88 km/h (55 mi/h) and still see a pedestrian in time to stop. (Most roadways with pedestrians have speed limits of 72 km/h (45 mi/h) or lower.) The results for target detection suggest that headlamp and overhead-lighting design might be inadequate for small-object detection.

Weber Contrast and Detection Distance

Pedestrian

Weber contrasts for the pedestrians and targets were calculated using the same procedure outlined in the scoping experiment. Weber contrast results were related to detection distances and are examined in the discussion.

Overhead Lighting at 100 Percent: The contrasts calculated from the images of the pedestrian are shown in figure 96. Overlaid on the contrast curves are points at which the participant detected the pedestrian. Between 152 and 213 m (500 and 700 ft), the curves slightly diverge, with pedestrians and headlamps on having more contrast. At those greater distances, the headlamps illuminate the roadway in front of the vehicle, causing the eye to adapt to the brighter foreground and causing the background and pedestrian to appear darker.

1 m = 3.3 ft

Figure 96. Graph. Overhead-lighting level experiment—pedestrian contrast and detection distance at 100-percent overhead lighting.

The contrast curves diverge more sharply at distances closer than 122 m (500 ft), where the headlamps begin to indirectly illuminate the pedestrian via reflections from the pavement and guardrails, and very sharply at 91.4 m (300 ft) from the pedestrian, where the headlamps begin to directly illuminate the pedestrian.

Changing contrast as the vehicle approaches the pedestrian did not appear to affect detection distance because the average detection distance with headlamps off was 285 m (925 ft) and with headlamps on was 250 m (819 ft), distances with very low measured luminance contrast. This level can be determined to be the threshold contrast for the object. (Note 100-percent detection threshold, not 50-percent detection as typically referenced.) The participants could be detecting pedestrians from farther away without headlamps because headlamps raise the ambient light level, causing the eye to adapt to a higher light level and be less sensitive. It would also be because participants tend to not look beyond the area their headlamps illuminate and do not scan as far ahead with headlamps on. Both detection distances are sufficient for a driver to stop a vehicle traveling at 121 km/h (75 mi/h) before arriving at the pedestrian; however, while it is unlikely a road with that speed limit would have pedestrians, other hazards, such as wildlife, might be present.

Overhead Lighting at 50 Percent: The Weber contrasts of the images were similar to those taken at 100-percent overhead-lighting level, because dimming the overhead lighting dims the entire scene and changes the luminance difference, while the luminance ratio is unchanged. Average detection distances were more than 244 m (800 ft) with 100-percent overhead lighting but are just over 213 m (700 ft) with 50-percent overhead lighting with headlamps on or off. A 50-percent reduction in overhead-lighting output shortened the detection distances by 21 percent without headlamps and 14 percent with headlamps. Results are shown in figure 97.

1 m = 3.3 ft

Figure 97. Graph. Overhead-lighting level experiment—pedestrian contrast and detection distances at 50-percent overhead lighting.

Overhead Lighting at 10 Percent:

The Weber contrast curves for 10-percent overhead lighting are similar to those for the other lighting levels, but the average detection distances are different (figure 98).

1 m = 3.3 ft

Figure 98. Graph. Overhead-lighting level experiment—pedestrian contrast and detection distances at 10-percent overhead lighting.

Target I (Low VI)

Weber contrast calculations were performed for the targets at 100-percent and 40-percent overhead lighting levels. The 40-percent overhead lighting level had considerably shorter detection distances than more intense levels and appears to be the level at which overhead lighting and headlamp light contribute similarly to luminance. Therefore, those lighting levels were selected for contrast analyses.

Overhead Lighting at 100 Percent:

Images of target I with 100-percent overhead lighting taken from various distances are shown in figure 99 to illustrate that, between 45.7 and 76.2 m (150and 250 ft), the contrast changes from positive to negative. At 61.0 m (200 ft), the 100‑percent overhead lighting creates negative contrast by illuminating the road behind the target, and the headlamps create positive contrast by illuminating the target. The target goes through a stage where it is invisible, and the contrast shifts from negative to positive and crosses the point of zero contrast.

Figure 99. Image Table. Overhead-lighting level experiment—images of target I at 100‑percent overhead lighting for from 15.2 to 123 m (50 to 400 ft) to the target.

The Weber contrasts for target I at 100-percent overhead lighting, with and without headlamps, are shown in figure 100. With the headlamps on (solid line), the shift from negative to positive contrast can be seen as the line passes through zero contrast. Detection distances, indicated by squares and diamonds on the two lines, are widely spread. When the vehicle was 76.2 to 122 m (250 to 400 ft) away, the contrasts, for headlamps off and on, like the luminances, were very similar.

1 m = 3.3 ft

Figure 100. Graph. Overhead-lighting level experiment—target I contrast and detection distance at 100-percent overhead lighting.

Overhead Lighting at 40 Percent:

Images of target I at various distances with 40-percent overhead lighting with and without headlamps are shown in figure 101. The ambient luminance is darker than at 100-percent overhead lighting. The same change from positive to negative contrast between 45.7 and 76.2 m (150 and 250 ft) is apparent in these images, as it was in the images with 100-percent overhead lighting.

Figure 101. Image Table. Overhead-lighting level experiment—images of target I at 40‑percent overhead lighting for from 15.2 to 123 m (50 to 400 ft) to the target.

Weber contrast and detection distances for target I in 40-percent overhead lighting are shown in figure 102. The average detection distance with headlamps on is shorter than with headlamps off, which could be because the headlamps cause the participant’s eye to adapt to higher luminance, reducing contrast sensitivity. Also, at 40-percent overhead lighting, because headlamp light dominates the visual environment, participants might be less likely to look beyond the area illuminated by them.

1 m = 3.3 ft

Figure 102. Graph. Overhead-lighting level experiment—target I contrast and detection distances versus distance from target at 40-percent overhead lighting.

Target II (High VI)

As with target I, the contrast of target II was analyzed at 100-percent and 40-percent overhead-lighting levels.

Overhead Lighting at 100 Percent:

The Weber contrasts for target II are shown in figure 103. With headlamps at distances closer than 76.2 m (250 ft), the headlamps create positive contrast on the target. Beyond that and without headlamps, contrasts are very close to zero but fluctuate owing to differences in the pavement behind the target and the changing visual angle.

Most participants detected the target more than 61.0 m (200 ft) away, showing that luminance contrast was not the only factor driving target visibility. The average detection distance without headlamps decreased by approximately 15 m (50 ft) (19.4 percent) when the lighting level was reduced from 100 percent to 40 percent. With headlamps, the detection distance decreased by approximately 12 m (40 ft), or 17.8 percent. Detection distances are nearly normally distributed when the headlamps are off. When the headlamps are on, the distribution is skewed toward longer distances.

1 m = 3.3 ft

Figure 103. Graph. Overhead-lighting level experiment—target II contrast and detection distances versus distance from target at 100-percent overhead lighting.

Overhead Lighting at 40 Percent:

Weber contrast at 40-percent overhead lighting, shown in figure 104, is similar to that at 100-percent overhead lighting. Without headlamps, the target detection distance is about 15.2 m (50 ft) farther away than with headlamps, even though the contrast with headlamps is higher. Again, this could be because headlamps increase ambient luminance causing the eye to adapt to brighter conditions and become less sensitive to contrast.

1 m = 3.3 ft

Figure 104. Graph. Overhead-lighting level experiment—target II contrast and detection distances versus distance from target at 40-percent overhead lighting.

Visibility Level

VL is a calculation describing the extent to which a target can be seen. It accounts for the object’s size, its contrast, the age of the observer, and the threshold luminance required for an object to be visible to 99 percent of viewers. The higher the VL, the more visible the target should be.⁽⁹⁶⁾

Pedestrian

The VL calculations for the pedestrian with headlamps on are shown in figure 105. Overhead-lighting levels from 10 to 100 percent were averaged (dashed line) because there was very little difference in VL between them. They are compared with VL for no overhead lighting (solid line). The pedestrian VL was higher for no overhead lighting at all distances up to 244 m (800ft). That roughly corresponds with experimental observations, where pedestrians were visible from slightly greater distances with no overhead lighting.

1 m = 3.3 ft

Figure 105. Graph. Overhead-lighting level experiment—pedestrian VL with headlamps on and with and without overhead lighting.

Target I (Low VI)

Figure 106 shows the VL calculations for target I with headlamps on, grouped by whether or not there was overhead lighting. The VLs are very similar at 30 m (100 ft) from the target and beyond. The VL rapidly increases at closer than 30 m (100 ft). That point is closer than the distance at which headlamp light falls on the target, about 76.2 m (250), showing that the VL calculation does not correspond with experimental observations.

1 m = 3.3 ft

Figure 106. Graph. Overhead-lighting level experiment—target I VL with headlamps on and with and without overhead lighting.

Target II (High VI)

The VL calculations do not align with experimental observations for target II. Target II’s contrast transitioned from negative to positive when the vehicle with headlamps on got close to it, but the VL calculations do not show that change (figure 107). At about 46 m (150 ft), the VL calculations also show a spike in visibility for no overhead lighting and a dip in visibility for overhead lighting. Those effects were not supported by the experimental data.

1 m = 3.3 ft

Figure 107. Graph. Overhead-lighting level experiment—target II VL with headlamps on and with and without overhead lighting.

Conclusions

A research objective for this experiment was to evaluate the impact of headlamps on driver visual performance. Results found that distant objects were more visible without headlamps than with headlamps, likely because the headlamp light caused the eye to adapt to the brighter environment, reducing contrast sensitivity. In addition, headlamps, which produced measurable amounts of VI up to 91.4 m (300 ft) away from a vehicle, affected object visibility at 213 m (700ft) away, likely owing to light scatter on the roadway and environment. The impact of this indirect light on visibility was negligible.

Evaluating how overhead-lighting level affected driver visual performance was the next research objective. High ambient luminance made a positively contrasting object more difficult to see, and low ambient luminance made a negatively contrasting object easier to see. An object with very low contrast would be difficult to see, regardless of object size and adaptation level; however, that poor visibility can be mitigated by adjusting object size and adaptation luminance. In addition, a target with negative contrast becomes briefly invisible as it is illuminated by headlamps and transitions to positive contrast. The effects of higher adaptation levels to brighter environments are particularly noticeable at night with low-contrast objects.

Other project objectives were to evaluate how headlamps and overhead lighting interacted to affect driver visual performance and to evaluate the minimum overhead-lighting level where target and pedestrian visibility was not compromised. This experiment found that headlamps combined with overhead lighting resulted in better object recognition and detail visibility than with overhead lighting alone, especially when overhead lighting was dim. The combination of headlamps and overhead lighting appears to have its greatest impact when the two sources contribute a nearly equal amount of lighting. This partial contribution can be achieved in two ways and was found primarily with respect to target visibility. The first way was when overhead lighting was dimmed to 30 percent (3 lx (0.28 fc), approximately 0.3 cd/m²(0.09 fL)) or below. Above 30 percent, the overhead lighting was typically the dominating light source, allowing for distant detections, especially for targets. Headlamps competed with overhead lighting when the level was below 30percent, resulting in effects on the properties of the target—contrast, adaptation luminance, and light angle—that affect visibility. The second way the partial combination was achieved was via proximity, depending on contrast polarity. In the intermediate zone (30.4 to 76.2 m (100 to 250 ft)), a negative contrast target must transition to positive contrast when headlamps become the primary source of lighting, causing it to cross a point of zero contrast. In this same zone, a positive contrast target experiences changes in VL and contrast that can affect visibility. In this situation, whether an object seen in positive contrast remains visible as a vehicle approaches depends on adaptation luminance. The impact of headlamps and overhead lighting was less pronounced in the case of pedestrians, primarily because of their angular size and nonuniform shape.

Additional results found that object size, ambient lighting level, and contrast affected the visibility of gray pedestrians and targets in this experiment. First, the larger an object, the more visible it was, as shown by the VL calculations; pedestrians had higher VLs than targets. Second, the ambient lighting level affected the eye’s adaptation, which in turn affected contrast sensitivity. Last, longer pedestrian-detection distances were measured without headlamps (and with overhead lighting), because this condition had a lower adaptation level, making a low-contrast pedestrian discernible beyond 422 m (800 ft). Orientation recognition, however, typically occurred only within 30 m (100 ft) of the object, requiring the driver to drive slower than 24 km/h (15 mi/h) to be able to stop in time to avoid colliding with the object.