Evaluation of The Impact of Spectral Power Distribution on Driver Performance

Executive Summary

Traditional roadway lighting uses high-pressure sodium (HPS) light sources, which provide high photometric efficacy. HPS light itself, however, is amber and does not render object color correctly. With the advent of light-emitting diode technology in roadway lighting, a new aspect of the light source is now being considered—its spectral power distribution (SPD). Broad-spectrum sources, with significant spectral output across the entire visible spectrum, potentially provide additional benefits to the driver; these light sources can provide better color information and can activate all of the photoreceptors in the eye more efficiently. This project investigates these effects and considers the potential benefits of a broad-spectrum light source.

This project is a comprehensive review of the applicability of mesopic factors to roadway-lighting applications particularly for higher speed roadways. These factors are based on the transition in the human eye from cone sensitivity to rod sensitivity and represent the potential benefit of broad-spectrum sources over traditional sources. While the model to determine the impact of mesopic adaptation to visual performance is well established and verified, both in the laboratory and in some very carefully prescribed experiments, the real-world applicability of the model has remained in question. Determining the impact of mesopic lighting on high-speed roadways is the focus of this effort.

In addition, a momentary peripheral illumination (MPI) system for highlighting pedestrians was developed and tested. The MPI system’s effect on visibility may have also been affected by the overhead lighting source’s SPD and level because pedestrians on the roadside might be detected in the periphery of the visual field, where mesopic effects occur.

This project was developed as a stepwise approach to the problems noted above. The first twosteps were to develop a scoping experiment that defined the nature of the effect of the SPD of overhead lighting on visibility and to provide guidance for development of the subsequent experiments. The primary outcomes from this scoping experiment were that both the type and level of overhead lighting significantly affected the detection and recognition of objects on the roadway. This was also evident for objects that were off of the roadway. One of the primary determinants for detection was the color of the pedestrian clothing and the targets in the roadway, thus indicating that color contrast is a significant component of object detection. The results also indicated that roadway lighting uniformity has an important role in object detection. The final aspect was that of headlamp color and intensity. In scenarios when overhead lighting was used, headlamp configuration did not affect visibility.

These results drove the direction of the next two experiments. The first was an investigation of conditions when headlamps have an impact on object detection and when they do not. The second was the investigation of the applicability of the mesopic model to roadway lighting.

In addition to spectral experiments that were performed, an investigation of the applicability of an MPI system was conducted to determine whether a system could be developed to leverage the spectral aspects of the visual response. Although the scoping experiment showed a minimal spectral effect of headlamp color on detection distance, two headlamp colors were used to further explore this relationship or headlamp color and the MPI. A mock-up MPI system was created with servo-activated headlamps that either tracked the pedestrians as the vehicle approached or highlighted them for a short time during which the vehicle approached. The results of this experiment were that the MPI system resulted in both shorter detection distances and an increase in detection rate. Headlamp color did not seem to have a significant impact on detection. When the MPI system highlighted an area across from a pedestrian, participants’ detection rates and distances for that pedestrian were lower. This highlights the importance of careful design of a full-featured MPI system; participants’ behavior indicated that they expected it to work properly, so it must not produce false positives, which could distract drivers from actual roadside hazards.

The next experiment was an investigation of the interaction of vehicle headlamps and overhead lighting on roadway-object detection. Small targets and a pedestrian were located in specific positions along the roadway that created high- and low-visibility conditions. The overhead lighting was then dimmed, and headlamps were turned off and on while participant drivers tried to detect the objects. Results indicated that the impact of the headlamps varied by object size. For most lighting levels, the overhead lighting was the dominant force driving object detection, but that was not the case when the overhead lighting was at the lowest levels. Headlamps were the driving factor for orientation-recognition distances—recognition of the direction the object was facing. The applicability of these results is critical for roadway lighting design. Headlamps dominate object recognition and also drive adaptation luminance. Therefore, the effect of the SPD of the roadway lighting may be overridden by the headlamps’ effect on adaptation level and the contribution of headlamp illumination to object luminance.

The other experiment resulting from the scoping experiment considered the mesopic model. Here both static and dynamic target-detection experiments allowed the research team to evaluate the mesopic model in the field. The static portion of the experiment was performed by determining the threshold contrast for small targets, and the dynamic portion examined target detection from a moving vehicle. The results indicate that overhead-lighting level significantly affected object detection; higher adaptation levels resulted in a lower threshold contrast. The results also showed that in the dynamic experiment, higher speeds typically resulted in longer detection distances. In terms of the mesopic model, for white overhead-lighting sources, the experimental results corresponded well to the model; however, for HPS sources, they did not. One of the important aspects of the project was the consideration of the off axis or object eccentric to the line of sight. An issue with the mesopic model could be that it does not include a term for eccentricity that accounts for different retinal sensitivities at different angles. The main conclusion was that, although the mesopic model predicted some of the results at lower lighting levels, it also had limitations.

The final experiment performed did not attempt to limit driver eye glances or fix eccentricities at detection. This experiment included an MPI system, overhead lighting, two speeds, and pedestrian detection in the periphery of the visual field. The results indicate that, for pedestrians close to the roadway, there was no impact of overhead lighting’s spectral distribution on detection distance. For those pedestrians, adaptation luminance was the most influential factor affecting visibility. For pedestrians farther from the roadway, spectral effects were more significant, but those results might not be applicable to roadway lighting design because objects that far away would have to be moving fairly quickly and on a collision path with the vehicle to become a hazard. The MPI performance results were similar to those of the initial MPI experiment. The MPI reduced detection distances and increased detection rates for objects in the periphery of the visual field.

The results of this experiment show that in a natural driving environment at the speeds tested, there is limited applicability of the mesopic model to lighting design. It is likely that drivers scan the roadway and detect objects in the fovea, where mesopic effects are not seen. Headlamps might also cause a higher adaptation luminance than predicted from the road luminance, further limiting the applicability of the mesopic model to lighting design for nighttime driving.

In general, the conclusions from the project are that the spectral component of the light source affects driver visual performance but only in certain conditions. The adaptation luminance of the driver is by far the dominant component of driver visual performance. The speed of the vehicle also affects the driver visual performance. In general, for high-speed roadways, it is recommended that spectral effects not be included in the design of the lighting systems. For lower speed roadways where the lighting system is predominantly for pedestrians, it is believed that spectral effects may still apply.