U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
The IESNA recently approved a revision to its publication RP-8, American National Standard Practice for Roadway Lighting. The revision includes three methods for designing continuous lighting systems for roadways: illuminance, luminance, and STV.
One of the primary reasons for conducting the European study was to meet with leading experts in the field of roadway lighting to find out about their experiences with using a visibility design metric.
While the panel found that a lot of research is being done in the area of visibility, none of the research has yet been implemented into everyday practice. In more than one country, team members heard the words, "We have no practical experience," when it came to applying the visibility design techniques.
Because of a negative experience, the Swiss have changed their approach to lighting crosswalks. They used to shine lights directly across the crosswalk, but discovered that, when at the curb, the pedestrian was less visible because the background varied, from buildings in some places to darkness in others. The Swiss now light crosswalks from the side, so the pedestrian is highlighted in positive contrast. Later input from the French confirmed the Swiss approach, but included the caveat that the main risk is that pedestrians often believe they are seen by drivers whatever the light distribution and weather conditions, even if they are not in the zebra marking.
The panel was shown roadways in Finland that appeared to have relatively nonuniform lighting. It was thought these lighting systems might provide a higher visibility level. Subsequent calculations by a team member have, indeed, shown that this road exceeded any luminance and STV requirements in the new ANSI/IESNA RP-8-00. Naturally, the ultimate measure of the quality of this type of design will be the change in the number of crashes. Crash data were not available at the time of the visit.
The panel was pleased to notice the amount of visibility research being done in both France and Belgium. Team members saw spatial frequency analysis by Fast Fourier Transforms being used by several people (Eric Dumont, Philippe Boogaerts, etc.) to describe information content of a scene or border (edge) contrast. Additionally, Mr. Boogaerts indicated that the Fast Fourier Transform is also used in the processing of the images of charge coupled device (CCD) cameras. Both countries have selected the visibility model and equations of Dr. Werner Adrian and are using three-dimensional targets. Representatives in both countries told the panel that the visibility concept provided a more complete approach to lighting design and supplemented the information provided by the luminance approach that is commonly used throughout Europe. The approach, which utilizes three-dimensional targets, results in very uniform lighting (see figures 9 and 10).
The use of three-dimensional targets by the French and Belgians provides contrasts within the target, thereby making the target more visible.
Figure 9. Uniform vs. nonuniform lighting.
Figure 10 shows a typical three-dimensional target used by the Belgians to develop simulation software. The Belgians 3 found good correlation between panel ratings and STV calculations for 20 percent targets, but stated that "the visual task of a driver cannot be considered as detection, within a useful time, of unexpected small static targets." They further stated, "The use of STV assumes full use of (factors) affecting visibility and knowledge of the limitations of the concept." They were adamant about the need to include headlights into the calculation of visibility.
Figure 10 shows a typical three-dimensional target
Based on extensive research done by Figure 9. Uniform vs. nonuniform lighting. Jacques Lecocq, the French have proposed that a simple minimum target visibility level (VL) metric is all that is necessary. Mr. Lecocq's research is based on translating a model roadway into a computer program that allowed many observer trials and the rapid collection of data (figure 11). The panel noted that Mr. Lecocq's model relied on approximations of key factors. These factors include the use of Lambertian distribution calculation of light reflected from the pavement and the shadow effect of multifaceted targets. Mr. Lecocq noted that, as targets get larger, the visibility always becomes greater. Large targets develop contrasts within themselves, as opposed to small targets, which are always viewed against their background, i.e., the roadway surface.
Figure 11. Model roadway installation.
The nine possible target positions are shown in figure 12, a view of Mr. Lecocq's software. The software permitted experimentation that determined the minimum visibility level needed for adequate lighting.
Figure 12. View of Lecocq's computer modeling software.
Based on an R2 roadway, a 0.35-s observation time, and a 20 percent reflectance target, the results of the study indicate that a minimum visibility level of 7 is needed for good visibility.
Using a ray-tracing computer program called "Radiance," the Belgians have developed synthesized computer targets that replicate real-world, illuminated, three-dimensional targets on a demonstration roadway, as shown in figures 13a, b, and c.
Figure 13a. Demonstration roadway with three dimensional spheres and square, flat targets.
Figure 13b. Photographic image (zoom) of targets.
Figure 13c. Synthesized image of targets.
Studies utilizing the synthesized images have shown excellent correlation between the calculated levels of visibility and the subject assessments of the observers for both flat 20-cm x 20-cm and spherical targets (appendix D).
In addition, the Belgians believe that their work shows that good uniformity on a poorly lighted (<1 cd/m2) road is insufficient. While the VL does improve as the roadway becomes more nonuniform, they believe that Belgian drivers would not accept the appearance of the roadway (figures 14 and 15). Later input from Mr. Lecocq further clarified that this increase in VL only applies to the average of several targets in the sense of mean values. Further, if one considers one target at a time whose reflection factor is variable, a flat one can be made visible or invisible simply by choosing an appropriate reflection factor. The flat target can even play the role of a type of specious amplifier for the average of individual VLs. This is generally not the case for a spherical target. On a roadway with poor longitudinal uniformity, typically all targets are either very visible or invisible depending on location. In this case, however, a corresponding mean value is not related to the ability to see any obstacle at any place on the road at a given time by the driver.
Figure 14. Synthesized configuration on road surface (R1 - q0 = 0.1).
Figure 15. Synthesized configuration on road surface (R4 - q0 = 0.1).
The Belgians 4 have found that headlights impact VL and should be included in calculations. Also, they believe that the VL approach is not usable in cluttered environments, i.e., environments with off-roadway sources, such as towns and residential areas. Therefore, the VL approach should be limited to the lighting of main roadways in rural areas.
Finally, the Belgians noted that, with the addition of the visibility design approach, lighting engineers are no longer limited to "producing luminance," but can also "produce visibility."
The Swiss recently enacted a law recognizing that the pedestrian has the right of way in a crosswalk. The initial result of the new law was an increase in vehicle-pedestrian crashes. If circumstances in Switzerland are similar to those in the United States, the vast majority of pedestrian fatalities occur after dark. The Swiss studied the crosswalks and have based new crosswalk-and roundabout-lighting recommendations on the visibility principle of highlighting objects in positive contrast. As shown in figure 16, poles are positioned so that pedestrians are seen in positive contrast, when light levels are below 2 cd/m2. No special pole positioning is required for light levels at or above 2 cd/m2. Installation of the new lighting resulted in a two-thirds reduction in pedestrian-vehicle crashes, but an increase in minor vehicle-vehicle crashes, typically "rear-enders," resulting from quick stops.
Figure 16. Lighting scheme for crosswalks, Switzerland.
|<< Previous||Contents||Next >>|