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-RD-03-082
Date: December 2003
Minimum Retroreflectivity Levels for Overhead Guide Signs and Street-Name Signs
PDF Version (837 KB)
PDF files can be viewed with the Acrobat® Reader®
CHAPTER 5. FIELD EVALUATION
The objective of the field evaluation was to determine the minimum luminance needed to read overhead and street-name signs (as a function of distance). As described in chapter 2, there is a wide range of research findings related to legibility luminance requirements. More precise minimum luminance values were needed to determine the retroreflectivity that will produce those luminance values. The retroreflectivity values that produce the minimum luminance values are the MR levels that will be used to generate recommendations.
To obtain the minimum luminance values, an experiment was designed that involved nighttime viewing of overhead and street-name signs. Essentially, drivers were positioned in a closed-course, real-world driving scenario and were asked to read different retroreflective signs. The luminance of the signs was controlled so that they were initially too dim to read and then the brightness (i.e., luminance) was systematically increased until the words were read correctly. The remainder of this chapter summarizes the experimental procedure and findings.
For the overhead sign testing, two words were shown simultaneously on each overhead sign. There are three advantages associated with this approach. First, overhead signs usually contain more than one word. Second, this approach increases the efficiency of the data collection procedure, allowing more data to be collected in a shorter amount of time. Finally, by using the two-word configuration proposed, the resolution of the findings was increased (the top word had different luminance than the bottom word). Similar to the real world, only one street-name sign was displayed at a time.
This research was based on the legibility of words rather than other visual testing icons such as the Landolt ring or grating patterns. Each word contained six letters. These words were "everyday" or common words and were not associated with the name of a city or destination. In all, 15 different words were used for the overhead signs. The words were developed for and used in another TxDOT-TTI study where both the legibility and the recognition distances of overhead signs were determined for various ages of drivers (luminance was not controlled in this study). The words included seven neutral words and eight words with both one ascender and one descender. Table 15 lists the words.
The street-name evaluations were conducted during the same session as the overhead signs, but not simultaneously. To avoid potential learning effects, the majority of the street-name signs were made with different test words than the overhead signs. The street-name sign words used are also listed in table 15.
Table 15 Test Words
All sign backgrounds and sign legends were fabricated with type III sheeting. The street-name signs were constructed with new type III sheeting that consistently measured approximately 320 cd/lx/m2 for the legend and 55 cd/lx/m2 for the background. The overhead signs were fabricated for another study that was conducted approximately 5 years ago; therefore, there was some loss of retroreflectivity for the words in the overhead signs. To determine the extent, each letter of each word was measured six times. The average scores ranged from 230 to 290 cd/lx/m2 and are shown in figure 3. The green overhead background measured 40 to 45 cd/lx/m2.
Figure 3. Overhead Sign Retroreflectivity Values
The overhead signs were made with white Series E (Modified) 16-inch uppercase and 12-inch lowercase words on a green background. The street-name signs were made with white Series C 6-inch uppercase words on a green background.
Spacing between letters was in accordance with the standard highway alphabet as recommended by FHWA. For the overhead signs, two words were shown on the overhead sign. The spacing between the words was 34 inches (see figure 4). Only one street-name sign at a time was shown.
Figure 4. Layout of Overhead Sign Panel and Legend
Using the literature review and current practices survey described in chapters 2 and 3, sign positions were selected to represent typical sign locations. The bottom of the overhead sign was positioned 18 ft above the road surface. Figure 4 illustrates the precise positioning of the test words. The bottom of the street-name sign was positioned 9.5 ft above the roadway surface. This height was selected to simulate the practice of installing street-name signs on the top of STOP signs.
The MR modeling research addressed lateral positioning issues associated with various viewing geometries and various headlamp profiles. For the field study, the overhead targets were centered above the travel lane and the left edge of the street-name sign words was mounted 6 ft to the right of the right edgeline of the travel lane.
The same vehicle was used throughout the entire data collection effort–a 2000 Ford Taurus, Model SE. The Taurus headlamps were the tungsten-halogen VOA style. Specifically, the driver's side headlamp was HB5 VOR LH DOT SAE AHRT5P2P 00T2 and the passenger's side headlamp was HB5 VOR RH DOT SAE AHRI5P2P 00T2. VOR means that the headlamp is to be visually/optically aimed using the right side of the cutoff, which is to be adjusted such that it is on the horizon line (at the same height as the center of the headlamp) when shown at a wall 25 ft away. In general, the VOA headlamp design (which includes VOR and VOL subclassifications) casts a relatively small amount of light above the horizon, not unlike the European headlamp specification.
SUPPLIED LUMINANCE LEVELS
Using both the low beams and the high beams, the researchers were able to provide 32 different, but precisely controlled, headlamp illumination levels to vary the luminance of the test words. The headlamp illuminance levels produced sign luminance values ranging from near zero (i.e., too dim to read) to that allowed by the maximum output with high beams (actual maximum sign luminance levels varied as the distance from the test signs varied). An attempt was made to control the headlamp illuminance levels so that the intervals producing sign luminance values near the standard threshold value of 3.4 cd/m2 would be small. However, as the headlamp illumination level increases, thereby increasing the sign luminance, the size of the intervals increased. A nearly constant legend:background luminance contrast ratio of 5:1 was maintained throughout the luminance range. Table 16 summarizes the luminance values that were supplied for each sign position. Figure 5 illustrates the luminance curves for each sign type and position.
Table 16. Supplied Legend Luminance Values (cd/m2)
Figure 5. Supplied Legend Luminance Graphs
Several methods of reducing the output of automobile headlamps are available. One method uses a variable resistor to dissipate a portion of the voltage as heat, with the remainder powering the headlamps. This would allow from 0- to 100-percent control of the light; however, the values in between would be nonlinear and would be difficult to replicate. Also, up to 100 watts (W) of power would need to be dissipated as heat. Another method used to control the light output is pulse-width modulation (PWM). This method applies full voltage to the headlamps at all times, but is interrupted at rapid and controllable rates. With the voltage turning on and off 2000 times per second, the ratio between the on-time and the off-time controls the brightness of the lamps. For example, if the voltage to the lamps was on for 50 microseconds (µs) and off for 450 µs, repetitively, the overall effect would be that the lamp is only receiving power for 10 percent of the time. This second method was chosen for this project.
Since we are now dealing with numbers, precise control of the light output is possible with a numeric processor or imbedded microcontroller. For this purpose, a Parallax BASIC Stamp 2 (BS2) was used. The BS2 contains a computer chip, serial input and output, 16 binary input/output lines, data storage, and memory. The BS2 is programmed with a standard laptop computer and retains the program until programmed again. To control headlamp output, a 16-position, binary rotary switch was used. The four-line output from the switch is sensed by the BS2 and, using a lookup table, produces the required PWM signal to the headlamp drivers. Since the percentage of on time does not easily equate to the percentage of light output as shown in figure 6, a switch position versus light output table was generated empirically with a laptop and a Tektronix J16 light meter and was programmed into the BS2. This method produces a highly repeatable set of test conditions than can easily be reprogrammed if necessary. The BS2, selector switch, and power switches are located in a small box that is held by the experimenter (figure 7).
Figure 6. Ford Taurus Headlamp Output
Special transistors were used to switch the headlamps on and off at 2000 times per second. These were power Metal-Oxide-Semiconductor Field Effect Transistors (MOSFETs), one for each headlamp. The common wire to each headlamp was cut and run to the drivers located on each fender. Since the common wire to the headlamp is normally connected to the plus side of the battery, a special "high side" driver circuit was used. By controlling the common wire to the headlamps, dimming is achieved on both the low and high beams. The internal resistance of these MOSFETs is very low (0.02 ohms), so there is little heat generated and there is very little voltage dropped across them, allowing nearly normal full voltage to the headlamps.
To allow operation of the vehicle at night without the controller turned on, a relay was added to each driver box. This relay, through the normally closed contacts, bridges across the power MOSFET to provide full voltage to the headlamp. This relay is actuated when power is applied at the control box, allowing the headlamp voltage to pass through the power MOSFET.
Finally, a solid-state 4-milliwatt (mW) red laser was powered from the control box through a switch. This laser, located in the vehicle's grill area and pointing forward, provided a means of vehicle (headlamp) alignment each time it is returned to the test course. Figure 8 shows a picture of the aiming laser.
The aiming laser was installed in the grill area of the test vehicle as shown in figure 9. Then the laser could be used to aim the vehicle as it was positioned for each evaluation. Figure 10 shows how the vehicle was aimed.
Figure 11 shows how the luminance values for each setting were measured. Using an LMT1009, the researchers measured the luminance of each sign position using 24-inch by 24-inch panels of white type III retroreflective material. A 24-inch square was needed to fill the aperture of the LMT at 640 ft using the 6-minute aperture. Very precise control was needed to accurately reproduce the luminance values from one night to another. For example, the researchers had to be in the same position (e.g., front seats), there could be no substantial difference in the weight distribution throughout the car (e.g., another observer in the backseat or substantial differences in fuel levels), and the contents of the trunk were removed. The headlamp lens and windshield were cleaned each night before the evaluations were begun. The researchers also kept the fuel topped off after each night of data collection. Also, it was important to keep the LMT at the same height for each reading.
The researchers also learned that the vehicle used during the evaluation would periodically run an engine fan. When the fan would start and quit, there was a moment of unstable luminance readings. However, the luminance readings would return to their previous state within 1 s of the fan either starting or quitting. The luminance change was so slight that only after many subject runs were the researchers able to notice it with their naked eyes and it did not appear to impact the subjects' evaluations of the legibility of the test words.
Sealed-beam halogen headlamps are generally known for having a substantial color shift phenomenon when the voltage is decreased from the standard operating voltage. However, the test vehicle used herein did not have sealed-beam headlamps and the voltage was not reduced. Still, the impact of the chosen method to vary luminance was not known. Consequently, before the researchers fully implemented the experimental plan, chromaticity and color temperature readings were taken to determine the color shift patterns of the Taurus headlamps (which were tungsten-halogen replacement bulbs). This was a critical issue since a substantial color shift would add severe confounding to the legibility analyses.
Figures 12 and 13 show the chromaticity shift from the brightest setting to the least bright setting using the Commission International d'Eclairage (CIE 1931 color space (ASTM E308). Figure 14 shows the corrected color temperature (CCT) shift. A Photo Research PR®-650 was used to take both the chromaticity and color temperature readings. Both of the trends were determined to be inconsequential and the procedure was implemented.
Figure 12. Chromaticity Color Shift (CIE, 1931)
Figure 13. Closeup Chromaticity Color Shift (CIE, 1931)
Thirty subjects were recruited from the Brazos Valley, TX, area using advertisements at local senior centers. Subjects received financial compensation of $30. Each driver was required to have a current Texas driver's license without nighttime restrictions. Table 17 lists the subject data.
Table 17. Subject Information
All 30 subjects were at least 55 years of age. Twelve were between ages 55 and 65. The remaining 18 were age 66 or older, with the oldest subject being 81 years of age.
Because legibility is a function of vision, the visual acuity of each test subject was measured using a standard Snellen eye chart at a distance of 20 ft. Two subjects had visual acuity better than 20/20. Nineteen subjects had visual acuity of 20/20 to 20/30. The remaining nine subjects had visual acuity greater than 20/30, but none had visual acuity worse than 20/40.
Contrast sensitivity tests were also conducted using a VisTech VCTS® contrast sensitivity chart at a distance of 3.1 m (10 ft). An advantage of using contrast sensitivity as an independent variable is that it provides a comprehensive measure of visual function across a range of sizes and contrasts that appear in the roadside environment. Only 7 of the 30 subjects were classified as having marginal contrast sensitivity. The remaining 23 were classified as having normal contrast sensitivity.
No external sign lighting (the type of lighting designed to illuminate overhead signs) was used in this experiment. This area in which the study was performed can be considered rural with low ambient light. No glare sources were present other than that produced from the instrument panel inside the vehicle, which was maintained at the highest setting throughout the experiment. All data were collected under dry conditions (i.e., no rain or dew on the signs).
The objective of the experimental plan was to determine the minimum luminance needed to read overhead and street-name signs at legibility indices ranging from 40 ft/inch to 20 ft/inch, in 10-ft/inch intervals. The minimum luminance was needed to accurately determine the MR.
Subjects participating in the study were asked to meet the researchers at Texas A&M University's Riverside Campus. Subjects were asked to wear corrective lenses if they normally wear them while driving.
Upon arriving at the Riverside Campus, the researchers explained the study in general terms and asked the subjects to sign an informed consent waiver. Once the waiver had been signed, the researchers evaluated the subjects' visual acuity and contrast sensitivity at normal indoor luminance levels. These activities occurred inside a building at the Riverside Campus where a room was set up to perform the visual assessments.
Upon completion of the vision tests, the subject drove the test vehicle to the testing area with a researcher in the passenger's seat guiding the subject (approximately 1 mi through the decommissioned air force base). Upon arrival, the researcher read the test instructions and conducted a trial run. This allowed the subject to develop a familiarity with the testing procedure and allowed his/her vision to approach complete adaptation to the darkness.
The testing began with overhead signing. The subject was asked to drive to a specified starting location 640 ft from the sign (legibility index = 40 ft/inch) while using the laser to aim the vehicle. After arriving at the first test location and putting the vehicle in park, the researcher took control of the headlamps using the control box. The headlamps were turned off and the first set of words was installed on the sign. The researcher turned the headlamps on using the lowest illumination setting. The subject was then asked to read the words. If the subject could not read both words correctly, the illumination level was increased one level and the subject was asked to read the words again. This procedure continued until the subject read both words correctly two consecutive times. At this point, the researchers asked the subject to move the test vehicle forward to the next specified testing location associated with a reduction of 10 ft/inch of legibility index (in this case, the distance would be 480 ft or 30 ft/inch). The headlamps were turned off and two new words were installed (the selection of the test words was performed randomly throughout the experiment). The increasing illumination procedure was repeated until the subject consecutively read both words correctly. This procedure was repeated for the specified distances corresponding to legibility indices of 40, 30, and 20 ft/inch. After all of the specified distances corresponding to all of the legibility indices had been tested, the complete procedure was repeated two more times (using a unique randomization of a 15-word set for each subject) to build repetition and thus decrease variability.
After the overhead signs were tested, the same procedure was used to evaluate street-name signs. The one difference was the specified distances associated with the legibility indices. The letter height on the street-name signs was 6 inches and therefore the testing distances were closer than for the overhead sign evaluation. The total evaluation time took about 90 minutes. Figure 15 shows an illustration of the test course. Figures 16 and 17 show pictures of the data collection stimuli for overhead and street-name signs, respectively.
Figure 15. Test Course
The researchers recorded the responses at each illumination level, regardless of whether the subject could read the word(s) or not. The researchers also recorded all errors that the subjects made in reading the words.
Once the subjects completed the legibility evaluation, they were escorted back to the vision testing room. The researchers then conducted a brief exit interview and paid the subject for his/her time.
To ensure experimental control, the researchers remeasured the supplied luminance values to verify the repeatability of the initial luminance readings and to ensure that nothing had changed during the evaluations. The readings provided the confidence that nothing had changed during the evaluations.
In other efforts to obtain the best experimental control possible, the test vehicle was dedicated exclusively to this project throughout the duration of the data collection activities. No other individual was permitted to use the vehicle. Furthermore, the test vehicle did not leave the research site. These precautions were implemented to avoid the possibility of anything happening to the vehicle that could have caused headlamp misalignment. In addition, every test subject who participated in the study received the same set of instructions. This included directions to not guess at the legibility of a word. Rather, subjects were asked only to respond when they were reasonably confident in their answer.
In all, 30 subjects completed the study. All but one subject read 18 overhead signs and 9 street-name signs. The one exception was that one subject only read 12 overhead signs (because of time constraints associated with the subject's personal schedule). In total, there were 534 overhead sign observations and 270 street-name sign observations.
The most efficient way to illustrate the resulting data is by cumulative distribution graphs showing how much luminance is needed to accommodate the various percentages of the study sample. Figures 18 and 19 show these cumulative distribution plots for overhead signs and street-name signs, respectively.
Using figures 18 and 19, it is relative easy to develop the luminance values needed to accommodate the various percentages of the study sample (at distances corresponding to the different legibility indices). Table 18 shows the results.
Table 18. Threshold Luminance Values by Accommodation Level (cd/m2)
1For white Series E (Modified) 16-inch uppercase and 12-inch lowercase words on a green background
2For white Series C 6-inch uppercase words on a green background
However, the data in table 18 are shown as discrete (categorized by the distance corresponding to the legibility index and the letter height) rather than continuous. To determine the minimum luminance at other distances, the data were plotted as a function of distance. This allows interpolation of any distance within the range studied, which corresponds to the legibility index and the letter height. For overhead signs, this corresponds to a range of 320 to 640 ft. For street-name signs, the range is from 120 to 320 ft. Figures 20 and 21 show the results.
Probably the most referenced publication related to minimum luminance was by Sivak and Olson, published in 1985. (26) Their work included the geometric means of various luminance studies that had been previously published. They assumed legibility indices of 50 and 40 ft/inch of letter height for younger and older drivers, respectively. Their recommended minimum luminance values are shown in table 19 with the 40-ft/inch results of this study.
Table 19. Replacement Luminance Values
The results from the Sivak and Olson work compare well to the findings found herein. For all three replacement levels, the Sivak and Olson luminance criteria fall between the overhead and street-name criteria found as a result of the field studies.
Interestingly, for street-name signs, the results of the study are generally higher than for overhead signs or what Sivak and Olson have recommended. One possible explanation of this was that many subjects repeatedly commented on the difficulty they had reading the street-name signs because of a perceived letter spacing that was too close. They also commented that the all-uppercase design of the street-name signs made it more difficult to read because of the similarity in the word footprints. Had the street-name signs been made with an initial uppercase letter followed by lowercase letters, the threshold luminance values may have been lower.