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-04-145
Date: December 2005
Enhanced Night Visibility Series, Volume XIV: Phase III—Study 2: Comparison of Near Infrared, Far Infrared, and Halogen Headlamps on Object Detection in Nighttime Rain
PDF Version (929 KB)
PDF files can be viewed with the Acrobat® Reader®
The glare produced by oncoming traffic and the subsequent visibility decrement for drivers at night is a primary concern of transportation research. Different VES designs improve forward visibility, but they may reduce visibility for oncoming drivers. HID headlamps have increased light output to levels much greater than conventional halogen headlamps, yet the implications of this increase for glare are still unclear. With so many vehicles using different headlamp types, the goal of this study was to compare a set of categorically different headlamp designs in relation to glare. Glare can be described both subjectively and objectively; therefore, this study was separated into the subjective discomfort glare portion and the objective disability glare portion. A comprehensive evaluation of different headlamp designs with respect to certain characteristics such as driver age and light adaptation would be valuable to designers in mitigating the effects of glare while maximizing driver visibility.
This chapter ties the disability and discomfort glare portions of the study together by looking at each headlamp design with an overall perspective on performance. This section also discusses the research questions that laid the foundation for this study, including specific factors directly related to oncoming glare associated with nighttime driving resulting from various types of vehicle headlamp designs.
As discussed in the Results section, something may have happened to the low/wide VES two-thirds of the way through the study that caused this VES to have substantially larger illumination at the driver’s eye point when detecting pedestrians on the left. To determine the potential influence of this VES on some of the following research questions, a secondary ANOVA was conducted with this VES eliminated from the models. This analysis resulted in no changes in the statistical differences for the dependent variables of detection distance and deBoer glare rating. Not surprisingly, this analysis did eliminate all the significant interactions for the illumination at the driver’s eye point with the exception of the Pedestrian by VES interaction. The results for this analysis are shown in appendix J. Where applicable, the impact of this VES’s potential change is discussed in the following answers to research questions.
What effect will different glare sources in terms of intensity and beam distribution (low/narrow, low/wide, medium/medium, high/narrow, and high/wide) have on the performance of the disability glare pedestrian detection task?
Before the specific relationships of each of the five glare VESs are discussed, it is important to determine the overall effect glare had on visibility, and therefore, the task of detection. That is, in addition to looking at the differences in detection distance under various glare conditions, it would be useful to compare these detection distances to a baseline or no-glare condition. This comparison would allow an estimate in pedestrian detection difference caused by glare. The ENV clear weather study (ENV Volume III) had a task similar to the detection of the pedestrian on the right but with no oncoming glare to degrade visibility. Factors such as participant demographics, experimental conditions, and roadway were comparable between the two studies. In addition, both studies used the same experimental vehicle with the same headlamps, halogen headlamps at a low profile (HLB–LP), so the data from the ENV clear weather study will be used as the baseline for this comparison. The static pedestrian on the right side of the road in the ENV clear weather study had a mean detection distance of 242.6 m (796 ft) using the HLB–LP headlamps. This static pedestrian can be compared to the right pedestrian in this glare study, which yielded a mean detection distance of 122.8 m (403 ft); therefore, the addition of an oncoming glare source led to approximately a 50-percent decrement in the visibility of the right-side pedestrian. This decrement confirms that glare can have a detrimental effect on driver performance.
The relationship between the individual glare VES source and detection distance shows that the halogen (low/narrow) VES allowed oncoming drivers the shortest mean detection distance among all the lighting designs. In other words, drivers would have the least time to react and stop when encountering an obstacle in the glare of these headlamps. In fact, the halogen headlamp detection distances were more than 26.8 m (88 ft) shorter than any other headlamp. These lamps also had the highest illuminance at the driver’s eye at the point of detection. The low/wide (HID) headlamps allowed a mean detection distance of more than 43.3 m (142 ft) farther than the halogen. A likely reason the low/narrow halogen headlamps were more glaring is the aiming procedure used in this study, which aimed headlamps higher and more to the left than typical; however, as shown in the ENV discomfort glare study (ENV Volume VII), the different aiming strategies did not elicit a difference in perceived glare. Another possible explanation for the low/narrow headlamp being more glaring is the lack of a distinct pattern and cutoff such as the HIDs. The halogen headlamp emits light in a less controlled pattern; therefore driving toward this type of beam may be more glaring because the perceived intensity of the light remains high throughout the approach. On the other hand, a set of HIDs may have a brighter, more distinct hotspot, but the precise beam pattern of the headlamp may reduce intensity toward the approaching driver, which would result in a lower perceived glare.
The higher-output HIDs, with both narrow- and wide-beam patterns, were perceived by the participants to be significantly more glaring than the low/wide and medium/medium HIDs. The participant’s detection performance reflected this finding, with both higher-output VESs underperforming the low/wide and medium/medium lamps. Therefore, both discomfort and disability glare seem to be affected mostly by the intensity or output of the headlamps. Both of the higher-intensity beams had an output of more than 40,000 cd, whereas the other, lower-intensity HIDs were no more than 31,000 cd. This would tend to indicate that headlamps with higher peak output also have higher intensity values throughout the beam pattern, resulting in greater intensity directed toward oncoming drivers. The horizontal and vertical angles from the glare sources to the oncoming driver ranged from approximately 0.58 deg (horizontal) and 0.063 deg (vertical) for the left glare source at 300 m (984 ft) to 2.39 deg (horizontal) and 0.189 deg (vertical) for the right glare source at 100 m (328 ft). A second factor that might affect disability glare could be the reflected light from the pavement. Higher-output headlamps will result in higher levels of illuminance due to this reflected light.
What effect will different light adaptation levels in terms of ambient lighting environment (low of 0.15 lx and high of 0.45 lx) have on the performance of the disability glare pedestrian detection task?
Two different light adaptation levels were set inside the vehicle to determine if pre-exposure to different light levels would make a driver more or less susceptible to glare. The two levels of driver illumination were 0.15 lx (low) and 0.45 lx (high). These levels are comparable to the levels of illumination a driver may encounter at night caused by different vehicle headlamps, vehicle interior lighting, and other changes in ambient illuminance. The results indicated no significant difference between the mean detection distance for high adaptation, 96.6 m (317 ft), and low adaptation, 94.1 m (309 ft). This lack of significant difference might be a result of the range of the high and low adaptation levels. When the participant vehicle approached the glare source, the eye was adapted to a lighting level based on the reflected light from the road surface, the interior lighting (in this case the adaptation light source) and the veiling luminance from the glare source. Both the interior illuminance and the road reflection remained constant, but the veiling luminance changed with the angle between the line of sight and the glare source. The average illuminance at the driver’s eye at the point of detection was 0.75 lx for the pedestrian on the right and 1.52 lx for the pedestrian on the left. This indicates that the glare source is the dominant source of illuminance at the participant’s eye at the point of detection, especially for the pedestrian on the left. It is possible that had a wider range of illuminance been tested, such as 0.75 lx, the adaptation level may have been significantly different; however, the values tested represent illuminance on real roadways in both lit and unlit conditions as measured during this investigation.
What effect will different pedestrian locations in the driving lane (left or right) have on the performance of the disability glare pedestrian detection task?
Two different pedestrian locations were used to evaluate the difference between the right and left sides of the driving lane. As expected, the right pedestrian location yielded a detection distance almost twice that of the left pedestrian location. One reason for this large disparity is the different angle of incidence as the driver approached the glare source. The left pedestrian was slightly to the right of the glare source. In fact, from the starting distance of 305 m (1,000 ft), the left pedestrian was in the same line of sight as the approaching glare source’s left headlamp. The right pedestrian was 3.66 m (12 ft) to the right, on the other side of the driving lane and, at the same starting distance, out of the direct line of sight of the glare source. Not surprisingly, the illumination at the eye of the driver was 50 percent less at the detection point for the right pedestrian as compared to the illumination at the point of detection for the left pedestrian.
What effect will different age levels (young (18 to 25 years old), middle (40 to 50 years old), and older (65 or more years)) have on the performance of the disability glare pedestrian detection task?
The pedestrian detection task involves many physiological mechanisms that are important in driver performance. As mentioned in the Introduction, driving a vehicle is primarily a visual task. The perception of and reaction to necessary visual information is even more important when driving on dark roadways at night with glare from oncoming vehicles. One crucial aspect of visibility in this difficult situation is the ability not only to track the course of the road but also to detect obstacles in the vehicle’s path. The same three age groups used in previous ENV studies were used to examine the effect age has on a driver’s ability to detect pedestrians in oncoming glare. The mean detection distances of each of the three age groups were statistically different. Not surprisingly, as the participant age increased, the mean detection distances decreased. Because each condition was conducted with oncoming glare, it is not known how much of this decrease was the result of disability glare being more troublesome for older drivers or how much was caused by older drivers’ deteriorated visual acuity. Previous research has shown not only a decrease in visual acuity with age, but also an increased sensitivity to glare.(4) Recall that the brightness acuity test was performed during the training to determine participants’ sensitivity to glare. The results of this test are shown in figure 20. In this figure, it is apparent that the older age group results were shifted toward a higher glare sensitivity than the younger group, indicating that disability glare likely played a role in this age-related decrement.
A comparison of these results to those of the ENV clear weather study (ENV Volume III) also indicates that disability glare had a larger effect on the older age group than on the other age groups. Recall that the clear weather study used the same VES and vehicle in a detection task similar to the disability glare study’s detection of the pedestrian on the right but without oncoming glare to degrade visibility. The three age groups detected the clear weather study’s static pedestrian on the right side of the road at the following mean detection distances: young, 257.6 m (845 ft); middle-aged, 222.7 m (731 ft); and older, 247.5 m (812 ft). When drivers were exposed to glare in the disability glare study, the mean detection distances of the three age groups were as follows: young, 143.2 m (470 ft); middle-aged, 121.8 m (400 ft); and older, 94.9 m (311 ft). This comparison shows a detection distance decrement due to glare for the young and middle-aged groups of approximately 45 percent and a detection distance decrement due to glare for the older group of approximately 62 percent.
What effect will different glare sources have in terms of intensity/beam distribution (low/narrow, low/wide, medium/medium, high/narrow, and high/wide) have on the perception of discomfort?
The five sets of glare headlamps not only significantly affected driver performance but also produced different perceptions of discomfort. The subjective deBoer scale was used to determine how driver comfort levels were influenced by the five glare sources. Perhaps the most interesting aspect of this study is how closely discomfort glare and disability glare were associated. For the most part, the more discomforting the VES, the worse the performance on pedestrian detection. As mentioned in the Introduction, research has demonstrated a subjective preference for viewing halogen headlamps over HIDs.(14) The increased brightness and bluish-white tint to the HID glare sources was perceived as more discomforting than conventional halogen lamps. Yet in this study, participants found the glare from the halogen (low/narrow) beam to be the most discomforting. These results match the objective detection distance results because the halogen beam also had the shortest detection distance.
The ratings of discomfort are an overall rating as the glare sources are approached. A likely reason the low/narrow halogens were more glaring is the aiming procedure used in this study; however, as shown in the ENV discomfort glare study (ENV Volume VII), the different aiming strategies did not elicit a difference in perceived glare. Another reason for this finding may be the dynamic aspect of this study; the headlamps were evaluated while the angle of incidence was changing. If the halogen headlamps, with a less confined beam pattern, continued to appear bright throughout the approach, then they might have been viewed as more glaring overall.
For the four HIDs, the beam intensity appeared to be more important than the beam pattern in causing discomfort glare. The two higher-intensity HIDs caused more discomfort glare than the less intense HID-based VESs. The fact that intensity level mostly determined discomfort level for all four HIDs demonstrates the drivers’ ability to identify the incremental differences between each headlamp. This relationship between maximum output of the glare source and perceived discomfort is further illustrated in figure 21; the blue dashed box represents the low- and medium-intensity VESs, and the orange dashed box represents the high-intensity VESs. The graph shows that the low/wide VES does not follow this same relationship of lower maximum output being associated with less glare; however, recall that there was an apparent increase in illuminance at the driver’s eye at the point of detection later in the study for this VES. A subsequent analysis was performed to look at the difference between the discomfort glare for this VES before and after this apparent increase. Before the apparent increase, drivers rated this VES as less glaring, with an average of 7.4 as compared to an average of 5.6 after the apparent increase. The 7.4 rating of the earlier participants is more in line with the expected discomfort rating caused by maximum intensity. Through this analysis, it is apparent that at some point during the performance of the study, the luminous intensity directed towards the driver from the low/wide headlamp changed. This most likely was caused by either an aiming or headlamp output issue as a result of damage to the headlamps that was not caught during the experimental process.
What effect will different light adaptation levels in terms of ambient lighting environment (low of 0.15 lx and high of 0.45 lx) have on the perception of discomfort glare?
Each discomfort glare rating was performed under both the low and high light adaptation levels. Adaptation level caused no significant difference in perceived glare. One reason there was no difference is likely because participants essentially were evaluating the addition of a glare source to their current adaptive state; therefore, whether the adaptation level was low or high to begin with, they still perceived the same difference in glare from the headlamps. As discussed previously, as the participant vehicle approached the glare source, the adaptation level remained constant and the veiling luminance from the opposing headlamps changed. As the driver approached the glare, illumination levels rose to anywhere from 1.5 lx to over 5 lx. These levels are much higher than the adaptation sources tested, and the participant likely evaluated the dominant source that changed across VESs, not the unchanging adaptation level.
What effect will different age levels (young, 18 to 25 years, middle, 40 to 50 years, and older, 65 or more years) have on the perception of discomfort glare?
The three age groups of drivers in this study had no significant differences in discomfort from the glare sources. In general, no matter what the age of the driver was, perception of the glare sources remained basically the same. These findings suggest that age may influence certain performance factors (e.g., object detection) because of changes in physiology, but it does not have an effect on subjective discomfort ratings. Multiple vision tests were administered to document the participants’ visual characteristics. One important physiological measure in this study was visual acuity, which gradually degrades with age.(25) A higher visual acuity may allow for greater detection distances, but it may not affect subjective ratings of discomfort. Another measure of visual acuity used in this study was an individual’s sensitivity to glare. According to Scheiber, older drivers are not only more limited with visual acuity but also may be more sensitive to glare.(4) An individual may have perfect visual acuity, but then the exposure to glare may reduce the ability to see. As discussed earlier, the BAT glare sensitivity tester used in this study measures the decrement in acuity caused by increasingly brighter glare sources. An increase in sensitivity would be represented by a decrease in visual acuity under glare (measured with a Snellen eye chart). For this particular test, it is unknown whether an increase in sensitivity would be represented by an increase in discomfort. It may be gathered, then, that a raised sensitivity to glare can lower detection distances, but it is unknown if it alters the feeling of discomfort. If declining visual acuity and glare sensitivity affect only pedestrian detection, then there would be no expected age effect for discomfort glare ratings.
What effect will different glare sources in terms of intensity/beam distribution (low/narrow, low/wide, medium/medium, high/narrow, and high/wide) have on the illuminance value at the driver’s eye at the moment of detection?
The concept of an illuminance reading at the moment of detection is interesting because it is related to detection distance. If a glare source has a low mean detection distance, then it would be predicted that that same glare source would have a higher illuminance value; the reading should be higher because in theory it is taken closer to the light source. The illuminance measurements taken at the moment of detection during the low/narrow trials (1.93 lx) followed this prediction, allowing a mean detection of only 66.8 m (219 ft); therefore, the lowest detection distance had the highest illuminance. The other four glare sources did not follow the same pattern. In fact, instead of falling into groups based on intensity, as detection distances did, the HID headlamps were separated more by beam distribution (narrow, medium, and wide). The low/wide and high/wide glare sources had similar mean illuminance readings of 1.44 lx and 1.31 lx. The other two HID sources (high/narrow and medium/medium) had mean illuminance values of 0.54 lx and 0.46 lx. This is somewhat expected. When drivers first approached a glare source, they perceived its maximum beam from a long distance with a very small angle between the driver and the source. As the approach continued, this angle increased, and the light reaching the driver actually came from the side of the beam rather than the end of the beam. If the beam was wider, more light went to the side, and a higher illuminance resulted; therefore, it is likely that the width of the beam has a greater influence on illuminance levels at the detection point than the maximum intensity of the glare source.
What effect will different light adaptation levels in terms of ambient lighting environment (low of 0.15 lx and high of 0.45 lx) have on the illuminance value at the driver’s eye at the moment of detection?
The results of this study showed no significant difference in driver’s eye illuminance levels at the moment of detection for light adaptation level. The results revealed similar mean illuminance measurements of 1.12 lx (low adaptation) and 1.15 lx (high adaptation) for the two adaptation levels.
What effect will different pedestrian locations in the driving lane (left and right) have on the illuminance value at the driver’s eye at the moment of detection?
As expected, the two pedestrian locations yielded two significantly different levels of illuminance at the moment of detection. The right pedestrians were detected nearly twice as far away as the left pedestrians. The mean illuminance value at the moment of detection for the left pedestrian was 1.52 lx. The right pedestrian was detected with a mean illuminance of half the left pedestrian’s, 0.75 lx. The main reason for this difference in illuminance is that the participants were much closer to the glare source when they detected the left pedestrian, and therefore, had a higher illuminance at the eye. These findings are consistent with a similar study reporting that the target closest to the glare source was very difficult to detect.(26)
What effect will different age levels (young (18 to 25 years old), middle (40 to 50 years old), and older (65 or more years)) have on the illuminance value at the driver’s eye at the moment of detection?
There were three age groups in this study and three significantly different levels for detection distance, yet there were only two significant levels of illumination at the eye. The older age group had a mean illuminance value at detection of 1.72 lx. The middle-aged and younger drivers had mean illuminance levels of 0.93 lx and 0.77 lx, respectively, which were not significantly different. The main reason for the difference in SNK groupings may involve visual acuity. The older drivers detected the pedestrians much closer to the glare source, and therefore, at higher illuminance levels. The middle-aged participants needed to be significantly closer than the younger drivers to detect the pedestrian. Still, both the middle and younger age groups were far enough away from the glare source that their illuminance values were not significantly different.
Few studies have evaluated disability glare caused by various HID headlamps in dynamic nighttime driving situations, and many questions concerning glare related to HID headlamps and other new types of vehicle lighting remain. This study can help guide future research to maximize forward visibility while minimizing glare. The findings in this study indicate that glare from oncoming vehicles can significantly reduce visibility even when pedestrians are dressed in light-colored clothing with a high reflectance of 40 percent. These findings differ from previous research that found that “higher reflectance targets [40 percent] are not significantly affected by headlamp glare even up to 5 lux at the eye.”(26) The current study found significant differences in detection distance with pedestrians in the right and left locations with driver’s eye illuminance readings less than 5 lx; therefore, new headlamp designs must be evaluated in terms of disability glare from oncoming vehicles before they are implemented into new automobiles.
The findings in this study led to the following guidelines for future consideration:
Many types of halogen and HID headlamps are in use on today’s roadways. Various headlamp designs have different characteristics that include intensity level, gradient or cutoff, and spectral power distribution. The collection of glare sources used in this particular study is only a sample of all the available headlamps on the market; therefore, certain generalizations about the performance of these particular halogen and HID designs may not be accurate when compared to other designs.
The drivers used in this study were given multiple training sessions before they completed the experimental tasks. In the disability glare portion of the study, the participants were familiar with the pedestrian types and roadway before beginning the data collection portion of the experiment. Participants were also expecting pedestrians to appear in the roadway, and therefore they were looking more attentively. Some of the results may have differed if unexpected objects and pedestrians had been introduced. In general, results from this portion of the study may not fully represent a real-world nighttime driving situation in which obstacles in the road need to be detected without warning. Drivers also received training before the discomfort glare portion of the study. The glare may have been rated differently if participants were not concentrating solely on the task of rating the glare. For example, if drivers drove as they normally would without being trained for the discomfort rating task, they may have not paid as much attention to the glare sources. This inattention may have led to smaller differences in discomfort ratings.
The illuminance measurements recorded during the discomfort and disability glare tasks were taken to measure the illuminance at the driver’s eye. The illuminance meter was mounted facing straight forward in the direction of the vehicle’s travel, and therefore, the meter did not account for driver head and eye movements. As drivers approached oncoming glare, they may have diverted their gaze to avoid looking directly into the light source. After analyzing the in-vehicle videos for eye and head movements, it was found that eyes and head positions moved for approximately one in three participants as the glare source was approached. The eye movements were mostly glances down or to the right side and increased in frequency as the glare became closer. Some drivers adjusted their head position either down or to the right in an attempt to mitigate the effects of the glare as they approached. The remaining drivers often had a fixed eye and head position throughout the detection task and looked straight ahead. One of the reasons drivers may have stared straight ahead more than they normally would is because they were in a searching mode. During the disability glare portion, the participants were primed for pedestrian detection, and the drivers were diligently fixating on the road ahead to detect pedestrians. The illuminance measurements may not fully represent the light reaching the eye of the participants who moved their head or eyes. It is also important to understand that the driver’s avoidance of a glare source is probably a normal response in a driving environment.
Many participants with similar visual acuity scores had differing sensitivities to glare. Although glare sensitivity was not a controlled factor, the vision test results revealed that no participant had a severe change in acuity resulting from the glare exposure (figure 20). In future studies, controlling for glare sensitivity may help researchers better understand the objective and subjective effects of glare and the different mechanisms involved.
The glare headlamps were stationary in this experiment to increase the safety and accuracy of data collection. For the same reasons, the experimental vehicle traveled at a speed of only 32 km/hr (20 mi/h). The results may have been different if both vehicles were dynamic and traveling at higher speeds.
Several avenues exist for future research concerning glare and night visibility. As technology continues to evolve in the area of automotive lighting, human factors research must keep up with the changing designs and ensure that the newest headlamps not only are more efficient and stylish but also provide better visibility and less glare. Certain technologies such as light emitting diodes (LEDs) are making their way into the market in other lighting applications and soon may be introduced as headlamp alternatives in production automobiles. A more comprehensive evaluation of multiple types of automotive lighting technology and various nighttime driving scenarios may be necessary to understand the overall effects of glare.
In addition, different roadway materials and roadway infrastructure characteristics may increase or decrease the perception of glare. For example, high barriers on busy interstates reduce the effects of glare by blocking some of the light, resulting in certain design implications for lighting and roadway engineers. In addition, overhead lighting may affect the perception of glare from different headlamps in various ways. A study that could test the interaction of both overhead lighting and new automotive lighting designs may be beneficial in further understanding the mechanisms behind glare.