Improvement of Visibility for Night Driving, With

IMPROVEMENT OF VISIBILITY FOR NIGHT DRIVING

Richard N. Schwab, Federal Highway Administration; and Roger H. Hemion, Southwest Research Institute

The research documented in this report confirms that at night on rural highways most motorists overdrive their headlights. Studies conducted to determine visibility of various highway vision targets and the effect of various lighting configurations on driver performance are summarized. Means by which night visibility may be improved are discussed. It is concluded that (a) a polarized headlighting system is the most promising system likely to solve the night visibility problem; (b) the system is technically and economically feasible in regard to today's vehicle popula­ tion; (c) the system would be advantageous in terms of improved visibility with less glare for motorists; and (d) the results of the use of such a system would be increased vehicular control, safety, and comfort and probably improved traffic flow and utilization of highways at night. A public test and evaluation of polarization in an isolated location is rec­ ommended prior to nationwide implementation to determine more precisely the benefits, side effects, and operational problems.

•FOR many years the Federal Highway Administration, through in-house and contract research, has been studying methods of improving the visual environment for night driving to provide more effective and safer use of our roadway network (1, 2, 3, 4, 5). Two methods are currently employed for illuminating the roadway at nighl-=ti.xed--=source "public" or "street" lighting and vehicular-mounted headlights. The fixed-source illu­ mination systems are comparatively expensive and therefore are usually considered applicable to the higher volume urban areas. At the present time only a small fraction of the total surfaced road and street mileage has some form of fixed illumination. Although urban driving now accoW1ts for approximately half of all travel and is in­ creasing its relative share, the absolute magnitude of travel on the other parts of the highway network is also increasing. One estimate suggests that the magnitude of travel on the two-lane rural highway system in the United states will increase about 50 percent within the next 20 years (6). Obviously, then, vehicular-moW1ted illumina­ tion must still be relied on for a large portion of night driving. The purpose of this paper is to summarize some 15 recent research and develop­ ment studies having two broad objectives: (a) to develop an understanding of the effects on traffic dynamics of glare and restriction in visibility during darkness, especially on two-lane rural highways and (b) to recommend for public trial a system of vehicular illumination that would be expected to provide an improved visual environment for night driving. The remainder of this paper will discuss the nature of the night visibility problem and then describe a series of studies designed to meet the two broad objectives stated. As a result of these studies, it would appear that a polarized headlighting system might be a promising solution. Therefore, past objections to the polarized system are out­ lined and means for overcoming them are suggested.

NATURE OF THE PROBLEM Headlighting design is currently based on a compromise between the need for adequate illumination of the road ahead and the need to avoid "dazzling" the eyes of the oncoming

Sponsored by Committee on Visibility and presented at the 50th Annual Meeting. 1 2 drivers with "glare" light. During the past half century a number of modifications have been made in the control of headlight beam configuration. These changes in in­ tensity, beam pattern, and aiming have improved the visual environment for night driv­ ing. Changes in beam configuration alone, however, cannot possibly provide adequate lighting to enable the driver to operate his vehicle safely during many common night driving situations because of the nature of the necessary design compromises. This is especially true on two-lane highways when meeting another vehicle. Because of the extremely small angular separation between the approaching vehicles and the variation in possible roadway geometry, it is necessary to limit radically the luminous output in the upper left quadrant of the headlight beam to keep glare out of the eyes of oncom­ ing motorists. Some luminance must be aimed in this direction, however, so that the driver can see things approaching from the left and can negotiate left-hand curves. Also, the designer is limited by physical considerations, such as filament size, in de­ termining how sharp this cutoff should be. Therefore, the possibilities of developing improved visual environments by beam configuration changes are extremely limited. The improved capability of seeing at night that comes with reduced glare can be expected to reduce the frequency of accidents. This is perhaps the major benefit. There are, however, many other aspects of nighttime driving that can be expected to benefit from better visibility of the road scene. Increased time will be available for driver decision-making because of the increased distances at which road obstacles can be detected. It should be possible to maintain safely somewhat higher driving speeds, with consequent increased traffic flow and greater utilization of highways dur­ ing off-peak nighttime hours. Although an improved headlighting system might produce an increase in night speed, this appears unlikely considering the small and directionally variable shifts in speeds encountered after installation of fixed illumination systems. As a result of elimination of glare, there would be a reduction in stress, a reduced sensation of "tunnel" driving with its pressures for more exact control of lateral posi­ tion, and an extension of the night driving capability of older drivers whose glare ad­ aptation responses have deteriorated.

BEAM USE STUDY To provirlP h::irkernnnrl nn thP PrlPnt nf thP. prnhlPm and data for benefit-cost analyses, a nationwide survey of headlight beam use was conducted (7) at 17 test sites in 15 states. Each site was 1,200 ft long and located, in most instances,-on rural two-lane unlighted highways of moderate traffic volumes. Three sites differed from these: Site 15 was a four-lane freeway with a median 50 ft wide; site 16 was a two-lane, high-volume rural road carrying a large proportion of recreational traffic; and site 17 was a suburban two-lane road with overhead lighting. One of the rural two-lane sites (site 7) was ob­ served under two conditions-snowy winter and clear spring weather. Observers at each end of the test site recorded the type of vehicle and headlight configuration on an event recorder chart with the vehicle position information. The instrumentation package automatically recorded time and position of the passage of individual vehicles in normal traffic on the same recorder chart. The observations showed that during normal nighttime, clear weather conditions over 75 percent of drivers were using their headlights improperly, i.e., using low beams when high beams should have been used because they were neither meeting another vehicle nor following closely behind one. To obtain sufficient data within a reasonable study time, the criterion employed in selecting sites was that they have an average daily traffic volume of 3,000 to 5,000 ve­ hicles. This results in minimum volumes at night of approximately 20 vehicles per hour. Therefore, the observed beam use patterns must be extrapolated to apply to lower volume roads not studied and prorated among all roads in order to estimate the beam use pattern on all rural highways. It was theorized that the average driver is not conscious of traffic volume as such but is aware of the time intervals between meetings with opposing vehicles. When this time interval is consistently "too short," the driver will no longer bother to switch back to high beam between meetings and will drive continuously on low beam. To 3 evaluate this theory and determine the effect of traffic volume on beam use when there was no other vehicle in sight, traffic volumes were recorded for each 15-min interval of the study. The intervals were classified by volume in increments of 10 vehicles per hour, and the percentage of high-beam use was plotted as shown in Figure 1. As can be seen, 50 percent high-beam use did not develop until the traffic volume had dropped to less than 30 vehicles per hour, i.e., when the driver sees an opposing ve­ hicle at intervals greater than 2 min. It should be noted that if a driver in moving traffic meets 30 vehicles per hour, the traffic volume counted at a fixed position along the roadway will be 15 vehicles per hour in one direction, provided that all traffic is uniformly distributed and moving at the same average speed. If traffic volumes in each direction are equal, the two-way traffic volume will be 30 vehicles per hour or the same volume as met by the moving observer. Thus, if these assumptions are met, the traffic volume in Figure 1 may be interpreted in two ways: (a) the number of opposing vehicles that a moving driver meets or (b} the two-way traffic volume counted at a fixed station. The confidence bands bounding the regression line as shown in Figure 1 indicate a wide variation in beam use from one observation to another at any specific traffic vol­ ume. However, Figure 1 does provide a reasonable method for making a projection because the mean values for all observations at any given volume grouping fall closely about the regression line. For benefit-cost and similar purposes, the mean value is of principal interest rather than beam use for any individual observation. The fraction of high-beam use for each volume grouping was multiplied by the fraction of nighttime travel occurring in that volume category. The result was a corrected estimate that approximately 39 percent of all unopposed driving at night in rural areas is done on high beams and the remaining 61 percent on low beams.

VISIBILITY STUDIES Because the greater part of rural night driving is done on low beams, it is important to determine the effect of both low- and high-beam headlights on the ability of drivers

100 ,---,-----,----,------.- ---.-----r-----r----.------,

- 0.0094834< Y= 72.5448e 90 \ \ 90% CONFIOENCE BANOS 801------1--- --1-- ~ ---l---- -1 OBTAINED USING THE TRANSFORMATION Jny= a+bx \ ~ 70 0 MEAN VALUE FOR EACH TRAFFIC w \ CD VOLUME GROUP r \ ~601--r --"'~---1- ----1-- - - ....i--- --4----~---~------~--1 0

10 ------

40 80 120 160 200 240 280 320 TRAFFIC VOLUME-VEHICLES PER HOUR

Figure 1. High-beam use as related to traffic volume. 4 to detect typical highway visual targets. A simulated roadway was established on a 5,000-ft asphalt runway of an inactive airfield (8). Two vehicles were instrumented to provide a continuous record of their position-both longitudinally and laterally on the simulated road, data on target detection by the driver, and glare intensity faced by the driver. From the position data, the distance between opposing vehicles could be de­ termined together with the distance of the driver from a target at the point when he detected the target. Variation of disability veiling brightness (DVB) with longitudinal distance between opposing vehicles with conventional headlights on two-, three-, and four-lane roads is shown in the upper curves of Figure 2. DVB is a photometric measure of the glare effect produced by all the luminances in the field of view. DVB is an expression of the equivalent veiling or uniform luminance that could be superimposed over some visual target to produce the same loss in visibility as that due to all of the glare sources in the field. Two characteristic effects resulting from increased lateral separation will be noted in Figure 2 by comparing the dashed curves for 4-lamp high beam on two-, three-, and four-lane highways: First is the reduction in intensity resulting from lat­ eral displacement of the observer from the center "hot spot" of the beam, and second is the movement of peak intensity to greater longitudinal distances between opposing vehicles.

POLARIZED HEADLIGHTING SYSTEM Several alternative methods have been suggested for reducing headlight glare. Many of the systems involve some type of screening or planting between vehicles moving in

5.0

2 LAMP HB 2LANE

vi I- a: 1.0 w ID ::;; I- ::::; m

200 400 600 800 1000 1200 1400 LONGITUDINAL DISTANCE BETWEEN OPPOSING VEHICLES, (FT) Figure 2. Headlight disability veiling brightness on two-, three-, and four-lane highways. 5 opposite directions. Application of these systems is limited to divided highways. Therefore, these systems cannot be employed on the very roads where they are most critically needed-the two-lane rural highway system. Another possibility is the use of controlled headlight beam distributions that optically project the beam straight ahead of the vehicle without impinging on the opposing lane. Unfortrmately, highways are not always straight, and rmless expensive feedback control systems are employed it is im­ possible to keep from blinding opposing drivers at some point in a curving highway situation. Of course, as was suggested earlier, a continuous fixed-source illumination system could be used, but this alternative is comparatively expensive. One of the most promising methods proposed as a solution to the night visibility problem involves the use of linear polarizers combined with high-wattage light sources (9, 10, 11, 12). Such a s ystem could theoretically provide an illumination s ystem withgreatlyincreased visibility of the r oad ahead without causing direct glare to op­ posing motorists . This is indicated in Figure 2 by the much lower level of DVB for the polarized systems (lower curves). The polarized system shown in Figure 3 r elies on the principle that two linear po­ larizers with their planes of polarization perpendicular to each other permit only a negligible amount of light flux to pass the second polarizer. Polarizers are attached to the headlights with their transmission axes oriented at 45 deg from the vertical. Another polarizer, called the analyzer, with the same transmission axes (i.e., parallel to those over the headlights) is installed in front of the driver's eyes in a position sim­ ilar to that of the sun visor but intersecting the line of sight toward the opposing ve­ hicle. This analyzer is constructed such that it can be inoved out of the way when no opposing vehicle is present. Because the transmission axes of the headlight polarizer and the analyzer are parallel, light from the driver's headlights that is reflected back from pedestrians, signs, pavement markings, and other parts of the roadside environ­ ment will be transmitted through the analyzer to the driver's eye. However, when another vehicle equipped in the same manner approaches, the transmission axes of the polarizers over its headlights will be perpendicular to that of the original vehicle's

OPPOSING­ VEHICLE'S

POl~*(J)LAMP(S)

~ -• ' ' I t I I I L~-~ f-1-1-/-l I I I I ~l ~ PEDESTRIAN

. ~ 0 POLARIZER

Figure 3. Polarized headlighting system : Light from the driver's own head lamps reflected from an object passes through the analyzer; light from opposing head lamps does not. 6

analyzer. Thus only negligible light is transmitted, and neither driver is blinded by glare from the approaching headlights. The main portion of the visibility studies involved determination of the distance ahead at which each of three targets first became visible employing four specific headlighting systems. The studies included (a) situations with no opposing vehicle (no other vehicle in sight) and (b) n,i.eetings of one opposing vehicle that employed the same headlighting system. In preliminary investigations using both one and those opposing vehicles, it had been determined that the increase in intensity and duration of glare caused by mul­ tiple opposing vehicles reduced detection distances; however, the number of cars in the platoon caused no differential effects among targets. Therefore the main experiment, employing 20 randomly selected drivers, studied only the "no-opposing" and "one­ opposing" vehicle situations. The following three targets were selected for this study as being typical of the range of driver detection tasks at night: 1. "Sign" target-a 96 percent diffuse reflective white 2¼-ft square with one quarter missing, seen against a 9 percent diffuse reflective black panel. The bottom of the panel was at a height of 60 in., and the left edge of the sign was located 6 ft from the edge line of the driving lane. The panel could be rotated and the missing portion of the square used as a target identification task. 2. "Pedestrian" target-a three-dimensional mannequin, 6 ft high, with exposed fea­ tures painted with 17 percent diffuse, reflective grey paint, wearing a 17 percent re­ flective grey topcoat, was used. The target was positioned with its right arm 2 ft from the edge line of the driving lane. 3. "Line" target-a portable, reflectorized, yellow, 4-in. wide, "no-passing" line 100 ft long. It was positioned in the normal location along the right side of the center­ line. The targets are shown on the right side of Figure 4. The results of these studies are summarized in Figure 4. As shown in the left curves, there was little difference in detection distance between conventional high-beam and low-beam headlighting with an opposing glare car that used the same headlighting mode, except for the sign target. When high beam was opposed by high-beam detection, dis­ tances for the sign were approximately 125 ft further away than when low beam was opposed by low beam. As expected, when an opposing vehicle was not present, detection distances were greater with high beam compared to low beam regardless of the target used. When unopposed, low beam provided detection distances that were at least as good as anything available to the driver in the meeting situation with high beam and usually were better. However, the visibility available to the driver in the meeting situa­ tion, with either high or low beams, was generally unsatisfactory and severely limited the time available for taking evasive action. What was clearly needed was a system allowing visibility equivalent to that of unopposed high beam during the meeting situation.

INCREASED DETECTION DISTANCES WITH POLARIZED SYSTEM When glare was radically reduced by use of polarization, the difference in detection distances between the opposed and unopposed driving modes was greatly reduced (Fig. 4). In the opposed mode, use of polarization increases detection distances over those ob­ tained in the unpolarized mode regardless of the target or the intensity of the lamp used with the polarization. In the unopposed mode, the increased detection distance will also be true if the analyzer is not in use (as it should not be when opposing headlights are not present) and if the polarized illumination is equal to or greater than that from the unpolarized standard headlights (criteria used in the design of the high-intensity polarized [HIP] system). Visibility without the analyzer and without an opposing vehicle would be equivalent to that of high beam for most types of targets. However, for the unopposed mode, detection distances shown in Figure 4 for the polarized headlight cases included the use of the analyzer because randomizing the presentation of the test factors to the subjects precluded the omission of the analyzer for unopposed trials. Also, this constitutes the situation that results if the driver becomes "lazy" and does not bother to move the analyzer out of his line of sight when opposing vehicles are not present. 7

··-

a. WITH OPPOSING GLARE CAR b. WITHOUT OPPOSING GLARE CAR -- LOW BEAM STANDARD HIGH BEAM.POLARIZED W/ANALYZER --• HIGH BEAM - HIGH INTENSITY, POLARIZED W/ANALYZER

Figure 4. Distribution of detection distances for all drivers and headlight configurations.

SUBSTANTIALLY HIGHER "SAFE DRIVING" SPEED An important question is bow the use of a polarized system affects the safety of the average motorist. Using the data on visibility of a pedestrian target at the road edge for each headlight system (Fig. 4), it is possible to calculate (13) a safe speed from which the normal (median) driver would be able to stop beforestriking the pedestrian. The results of this calculation are given in Table 1. The average speed in the sample of over 50,000 vehicles observed during the beam use study was 60.5 mph on dry pavements and 55,4 mph on wet pavements. It can thus be seen that the great majority of the drivers of these vehicles-approaching 98 percent under some road conditions-were exceeding a speed from which they could safely stop if a pedestrian stepped out onto the road. Pedestrians account for approximately 10 percent of the fatalities on rural highways. In the absence of an opposing glare car, there is a minor reduction in the safe speed in going from high beam to high-intensity polarized headlights with analyzer. However, this reduction can be compensated for by not using the analyzer when there is no opposing vehicle. The major effect is ob­ served in the case of opposing traffic, where the elimination of glare (disability veiling) inherent in the polarized system results in a marked increase in "safe speed" levels. 8

TABLE 1 SAFE SPEEDS FOR STOPPING (CONVENTIONAL VERSUS POLARIZED HEADLIGHTS)

Safe Speed (mph) Condition High-Intensity Conventional Headlights Polarized Headlights Opposing Pavement Vehicle Low Beam High Beam With Without Analyzer Analyzer

Yes Dry 40 40 63 Yes Wet 35 35 50 No Dry 50 80 73' 80 No Wet 40 65 58' 68

8 Not applicable- polarized system requires use of analyzer when meeting opposing vehicle; if not in use, it is the same as high beam. bWith analyzer in use-if the driver docs not use analyzer when there is no opposing vehicle, safe speed would be as indicated in "without analyzer" column

The foregoing discussion is based on the use of high-intensity polarized lamps. The use of existing high-beam lamps with polarizers would result in "safe speed" levels about halfway between those of conventional headlights and the high-intensity lights when an opposing vehicle is present. Without an opposing vehicle, the "safe speed" would be less than conventional high beam but well above that of low beam. Overall, use of existing high-beam lamps with polarizers (no increase in lamp power) would improve the night driving environment over that of conventional headlights in all cases except for unopposed driving where high-beam lamps are now used-39 percent of the unopposed driving. However, the improvement would not be sufficient to raise the "safe speed" for meeting situations to the level required to see and stop for a pedestrian at current highway speeds.

DIRTY WINDSHIELDS AND POLARIZER MISALIGNMENT Dctcctiu1J. capa.tility fur di·ivc:rs usi11g vula.iiZi:U li~ht~ will Lt:: .rt:a..iu\;~u Uy a tii1·i.y windshield not only because of reduced light transmission but also because the dirt particles are illuminated by the opposing headlights and interpose a more or less bright screen that effectively reduces the contrast between the tar get and its background. No appreciable depolarization was noted from the dirty windshield (8). The visibility loss from a dirty windshield when using the high-intensity polarized iamps was about 6 per­ cent (620 ft versus 660 ft) compared to 10 percent (530 ft versus 590 ft) with conven­ tional headlights. This result does not imply that driving with a dirty windshield should be ignored. This study only involved the effect of the dirt on detection capability, and the problems related to the distraction and confusion caused by dirt on the windshield were not considered. Some misalignment of the polarizers between the two approaching vehicles is bound to exist because of road crown, superelevated curves, and other causes of vehicular roll movement including rough pavements and potholes. Misalignment will increase the light transmission because of the nature of polarization. Dynamic studies of ve­ hicular roll have shown that a practical maximum of 7 deg occurs in moving vehicles on normal paved highways. As shown in Figure 5, a misalignment of more than twice this value, or 15 deg, has an insignificant effect on the driver's detection capabilities (14). It is therefore concluded that superelevated curves and other causes of vehicular roll will not adversely affect a polarized headlighting system.

COMFORT The discomfort occasioned by glaring headlights is essentially subjective and not necessarily accompanied by corresponding visual disability. Visual discomfort was therefore rated on a 6-point scale by asking the subjects, at the completion of each test 9

1,300,------,

E,,2001 •-=------..------~-..... '-....,HIP " w ~ 1,100- /'! (fl 0 z g 1,000- u w 1- w 0 ....--o HB't soo,,...------e------.,,,,,.-.!:'.:....--e

I I I 5 10 15 ROL L ANG LE

Figure 5. Mean target detection distances as affected by vehicular roll.

run, for a judgment of comparative discomfort caus ed by opposing headlights when meeting another vehicle head-on (8). As may be seen in Figure 6, there was definite correlation between subjective ratings of glare and measured disability glare levels. Both polarized systems produced only about one-third the discomfort (2 points less) of high beam and about one-half (1 point less) that of low beam.

FATIGUE STUDIES Reduction in drivi ng tension and physiological and psychological irritation of meet­ ing headlights of approaching traffic in nighttime driving, inherent in the elimination of glare, was subjectively apparent to all of the test subjects. It was postulated that driving a vehicle for extended periods, when all opposing traffic was controlled with regard to headlight modes, would induce fatigue at differential rates depending on the stressful

BLINDING 6.0 CJ "NO PASSING" LINE D PEDESTRIAN 5.0 -SIGN

"'z UNCOMFORTABLE \i 4.0 a: a:I- 0 ~ 3.0 0u (/l a BOTHERSOME 2.0

1.0

NO PROBLEM 0 HB LB P HIP LIGHTING MOD E

Figure 6. Relative discomfort from opposing headlight modes. 10

characteristics of the specific lighting modes, It was further expected from studies of other investigators (15) that sufficiently sensitive indicators of such fatigue propaga­ tion were available to discriminate among the comparative fatigue- or stress-inducing potentials of the various headlighting modes. Initially, the course used for the visibility studies was extended by utilizing an ad­ jacent parallel runway and an intersecting runway and taxiway to provide a loop course approximately 13/.i miles long. Real and simulated vehicles were utilized to provide opposing traffic (headlights) for the subject drivers. The development of fatigue in the observers was evaluated in terms of their performance of several physiological and psychophysiological tasks presented at the beginning and end of each test session or periodically during the session. The second portion of the fatigue study utilized simulated driving as well as road simulation. The instrumentation associated with this work was of laboratory type for the most part and provided evaluation of subject performance in three areas of mea­ surement plus a subjective rating response. These were areas of psychophysiological, physiological, and vigilance responses. The subject was placed in the driving seat of a stationary automobile. A tracking task was presented in the form of a meter "null­ ing" response to random input steering offset error signals. Random presentations of visual flicker and audio flutter required responses of lifting the right foot off the ac­ celerator and braking. These tasks were conducted in the presence of the repetitive presentation of the headlight modes being studied and each driver's performance measured. These experiments, in both the real driving and simulated driving phases, neither supported nor refuted the hypothesis that fatigue affected driver performance differently with respect to the lighting system involved. Although some of the measures and tech­ niques selected to indicate stress or fatigue development showed changes during the tests, in general, these indications were not consistent among drivers nor within a single driver's performance of successive replications (15). One response that did show some consistency was that involving reaction time in brake operation in response to an un­ anticipated signal. Figure 7 shows that no appreciable changes in\performance oc­ curred for the first 2 hours of the simulated driving session. However, in the third hour. differences were apparently related not so much to glare but perhaps to discom­ fort and stimulation. The polarized system apparently developed an optimum stimulus to alert the driver between the "overloading" effect of the glaring high-beam lights and the soporific effect of no opposing traffic. By using scaling technique, subjective discrimination of fatigue development related to lighting modes was achieved as shown in Figure 8. The greatest increase in fatigue was obtained in the absence of glare, and the least was obtained with the polarized sys­ tem, although the differences were not statistically significant. Thus only an indication of correlation with performance data was obtained.

DRIVER BEHAVIOR Another area of comparative interest investigated was the effect on the lateral posi­ tioning of the subject's vehicle in its lane when approaching the target and the meeting point with the opposing glare car(s). A consistent tendency to steer toward the oncom­ ing glare car and a distinct reduction in steering deviation was observed with most drivers at an intervehicular separation distance of around 800 ft (5 sec from meeting point). This can be attributed to an anticipatory "steadying" or "tightening" of steering control before the approaching encounter and a sort of "staking out of one's territory." Prior to this point, the approaching vehicle does not present a hazard; i.e., sufficient time is available to maneuver in the event something unexpected occurs. However, the presence of the opposing vehicle forces the driver to recognize that he is approaching a potentially hazardous situation and, therefore, he makes corrections to his vehicle's path to avoid a collision. This correction involves steering his vehicle away from the oncoming vehicle to obtain greater lateral separation. It was hypothesized that the greater stress of opposing high-beam lights would cause proportionately greater lateral displacement than reduced glare from low-beam or 20 X HB

NO GLARE ui 0 z 0u w Cf) w ::;c ;:::: z 0 t; 10

0 1..,.______....,______, ______......

0 2 3 HOURS OF ORIVING IN SIMULATOR

Figure 7. Effect of lighting conditions on reaction time at each hour.

HIGHER SCORE INDICATES GREATER FATIGUE 16 ,-- - --'\A,------, D BEFORE DRIVING 1·.::.;-::) AFTER 3 HOURS y 14 - / / ,," //~?

6 - L.r--...... __,,,-~------==-~~~ \{; );/: 0 O .____._N_O_.G..,L..,AR• E.,__ _.___H.a:B...:...... 1-....,__ _H ",I P,....___._ __, ~ 200 o 200 INTERCEPT DISTANCE, ( FEET)

Figure 8. Subjective fatigue mean score before and Figure 9. Maximum lateral deviation as affected by after simulating driving at each lighting condition. headlight beam type. 12 polarized headlights. There was some difference noted with regard to headlight modes in opposed trials (Fig. 9). The presence of the opposing vehicle itself appeared to be the principal motivating influence, irrespective of its lighting, causing the subject driver to move to the right as the vehicles approached their meeting point. Only in the un­ opposed trials did the target appear to exert any influence, but with essentially no dis­ tinction between the targets. Discrete judgments of such items as speed, distance, angular relationship, and the relative position of potentially hazardous situations are of importance to the driver. His ability to make these judgments may be influenced directly by the lighting system used. Particularly critical in this respect is the driver's judgment of when it is safe to leave a side road and proceed across a highway. studies were made of right-angle gap acceptance behavior using both conventional headlights and the polarized system (16). A recent re-analysis of the study results is shown in Figure 10. --iieadlighting systems that in the right-angle situation produced the most glare re­ quired somewhat longer gaps for crossing. Low beam and polarized beam with glasses were the "low-glare" situations requiring somewhat shorter gaps for crossing. Polar­ ized beam with visor and high beam were "high-glare" situations and required a larger gap. The variance in the minimum gap sizes accepted appeared to be about the same in all cases That the polarized system employed a fixed visor mounted in place of the sun visor should be characterized as "high glare" is reasonable in the right-angle sit­ uation because it does not provide the driver with protection against the vehicle ap­ proaching from the side. An analyzer in the form of glasses did provide protection to the side. However, because the head tips somewhat as it is turned to the side, the protection is not complete and the end result was quite similar to that of low beam. The high-glare conditions appeared to make drivers behave more conservatively by judging the brighter sources to be somewhat closer than they were and thereby allowing a somewhat greater margin of safety. This, of course, could have an adverse effect on traffic flow, particularly when a high volume exists on the main road. Fortunately nighttime traffic is usually not near capacity, particularly for most unlighted roads where the glare problem is especially acute. The gap acceptance studies clearly show that, if a polarized headlighting system is introduced to obtain the considerably superior

IOO r------,

0------0 LOW BEAM • - - • ~9M~E~~s8llM BO O •••••• O POLARIZED BEAM WITH VISOR X--X HIGH BEAM

a: w 2: 60 a: 0 LL 0 1- z tJ40 - a: w n. LOW GLARE ~ ,...... ,_...,. SYSTEMS

20

Q '------~10~0------20~0------3~00______. 400

GAP SIZE ( FEET)

Figure 10. Distribution of gap acceptance for each lighting mode. 13 forward visibility found with this system, care should be taken in the design of the ana­ lyzer to protect drivers during encounters with vehicles from the side. The design of the analyzer may be particularly important for older drivers, because the glare tolerance of individuals is increasingly reduced with age. Drivers over 65 years of age are seriously handicapped at night, and indeed they voluntarily and sub­ stantially reduce their night driving mileage.

FIXED LIGHTING The detection-distance results from low-beam headlighting alone (17) are compared with those obtained with the addition of fixed overhead lighting in Figure 11. As can be seen, the major effect of fixed lighting was confined largely to the two vertical tar­ gets. The reflectorized no-passing line was relatively unaffected by the additional illumination, because the brightness of the line was largely controlled by the illumina­ tion coming from the headlights. Because of the geometric relationship of line and driver, additional illumination from the fixed lighting sources was not effective in in­ creasing the contrast between the line and pavement. In the case of the two vertical targets-the nonreflectorized sign and pedestrian-the addition of fixed lighting to low­ beam headlighting more than doubled the detection distance. This appears to be pri­ marily due to illumination falling directly on the two vertical targets as will be dis­ cussed later. There was apparently little difference in the effectiveness of the two illumination levels employed-0.6 and 2.0 footcandles average horizontal illumination-as shown in Figure 11. The detection distance was also largely independent of the type of vehicular lighting used (Table 2). Glare from an opposing vehicle's headlights caused a reduction in detection distances, particularly when high beams were used, with or without a fixed lighting system. A much greater effect was observed with respect to the position of the target in rela­ tion to the overhead luminaires, i.e., whether the target was forelighted or backlighted. As is shown in Figure 12, front lighting increased detection distance for the two vertical targets. (The line target, being unaffected by fixed lighting, is not shown.) The in­ crease varied between 30 and 140 percent with an overall average of 80 percent. The differences were somewhat greater with the two-dimensional sign target than the three­ dimensional pedestrian, where the possibility of edge lighting was present. The loca­ tion of the luminaire is important and should be considered by highway lighting de­ signers when locating luminaires with respect to pedestrian crosswalks. In many

2000 ~------..------,

Figure 11. Detection distance with various levels of fixed lighting in addition to low-beam headlights on a dry pavement. 14

TABLE EFFECT OF VAR!OUS F!XED ANn VRH TrTTT.AH LJn H 'T'TNr,. SV.~'T'F.M C Ol\lTRTNA'T'TnNS ON MF.AN nF.'T'Rl"T'TON DISTANCE (IN FEET)

Fb

Dry Pavement

Unopposed Sign 420 865 BOO 1,467 1,360 1,710 1,367 2,092 1,563 1,960 1,783 Pedestrian 375 780 590 1,317 1,789 1,564 1,785 1,662 1,762 1,722 1,919 Line 215 360 380 177 175 249 290 336 411 582 497 Opposed Sign 345 490 630 1,417 1,330 1,191 1,122 1,536 1,260 1,005 1,377 Pedestrian 245 245 460 1,743 1,206 860 1,359 1,618 1,353 845 1,205 Line 130 135 260 158 197 153 238 340 267 184 424

Wet Pavement

Unopposed Sign 680 653 2,221 1,886 2,220 1,657 Pedestrian 2,298 1,695 2,008 1,280 Line 178 209 236 128 Opposed Sign 563 1,671 1,387 1,160 1,315 Pedestrian 1,722 1,560 991 1, 245 Line 182 160 102 172

Note: No data taken for blank cells.

instances, the two-way nature of traffic and other considerations make the provision of front lighting impossible. However, on one-way streets and/or at mid-block pedestrian crossings, for example, front lighting may be feasible. The location of the luminaire had the least effect when the driver employed parking lamps. This is reasonable because the target in the case of rear lighting is seen by silhm1P.ttP. :md any illumination reaching the target from the vehicular lighting systfim will only reduce the contrast. This may indicate that with adequate overhead lighting levels, motorists may be able to operate safely with only marker lights (perhaps of

RE AR FRONf -·~ ~l-

E 1500 w u FRONT LIGHTED z •

0 L....l-'-"1"""...l:.,.a,;(<',l..i...."""2..1:....ti~---..J..'".:.l~=L.L;;~:;a..,_J.Z/I..L...:l=------.....J HIP PARK LB HB HIP PARK LB HB SIGN PEDESTRIAN

Figure 12. Effect of target position with respect to fixed overhead luminaire. 15 somewhat large size and intensity) on their vehicles. This would reduce glare to a minimum. However, further investigation is needed relating to the ability of pedestri­ ans and drivers in detecting vehicles employing only marker lamps. Some research in this area is currently under way at Franklin Institute Research Laboratories in Philadelphia. Elimination of glare by use of polarization, by expanding the use of overhead fixed lighting, or by any other practicable means would greatly improve the ability of drivers to see roadside obstacles, traffic markers, signs, and other objects in the highway scene. However, the best of these alternatives from the standpoint of visibility appears to be that provided by adequate levels of fixed overhead lighting. As of 1967, there were 1.1 million miles of unlighted paved rural highways. As­ suming that traffic is insufficient to warrant lighting of unpaved highways throughout the country and that present headlighting systems could be utilized thereon, the estimated cost for installing and operating fixed lighting at a 2-footcandle average level for the paved road network for a 20-year replacement cycle is $87 billion (18). The total cost for fixed lighting would, therefore, approximate the total cost of thelnterstate Highway System. It is clear that the latter solution is not economically feasible for the entire highway system although it is justifiable on certain highways. Fixed lighting would appear to be warranted where the driver is faced with unpredictable situations in which the added visibility obtainable from fixed lighting is of considerable aid. Some factors that may lead to this lack of predictability are heavy concentrations of pedestrians, unusual ge­ ometry, and high traffic volumes. On unlighted rural roads, normal headlighting would still cause glare, and such roads constitute nearly all of the nation's highway system and carry the largest portion of the traffic. Therefore, a considerable increase in the application of overhead fixed high­ way lighting does not eliminate the need for modification of vehicle headlighting if the objective is adequately improved illumination for operation on the highway at night. In­ creased headlight intensity to provide greater visibility distances while eliminating glare from opposing headlights remains essential for the bulk of highway operations. Therefore, the polarized headlighting system is essential and has the added advantage of providing substantial improvement in visibility, coupled with a radical reduction in glare on all roads instead of the partial solution provided by fixed lighting employed only on certain roads.

OBJECTIONS TO A POLARIZED SYSTEM Numerous objections that have been raised against the adoption of a polarized head­ lighting system are no longer valid because of evolutionary changes in the motor ve­ hicle. One example is the use of laminated glass in the windshield, which eliminates distortions caused by tempered glass. Another example is the higher capacity of modern electrical systems, which makes higher powered lamps feasible. Some limitations still exist that must be further studied and overcome. Chief among these are the re­ duction of heat generated in the polarizers from the lamp and the transition to a po­ larized system from the present conventional lamps. These limitations are discussed in the following paragraphs.

Heat Problems Commercially available polarizing materials (and agents used for bonding them to the lamp) are currently able to withstand approximately 150 C without detrimental ef­ fects that reduce the transmission of light. The maximum power consumption obtain­ able for filaments designed to fit a standard 5¾-in. lamp appears to be about 125 watts. To obtain this limit, a dichroic reflector must be employed to dissipate the heat. Such reflectors are comparatively expensive at the present time. The lamp intensities re­ quired to obtain the increased detection distances shown in Figure 4 could be provided by four 50- to 75-watt lamps. Standard aluminum on glass reflectors could be used on such a lamp. The major problem of heat dissipation, therefore, would not arise unless longer detection distances than those shown in Figure 4 were required for very high 16 speeds. Because of the constraints imposed by vehicle power generation systems, it is not likely that heat dissipation will be a problem for any retrofit system. On new vehicles, if the heat problem can be solved, increased detection distances from very high intensity lamps might be considered.

Transition to the Polarized System The problem of conversion to a polarized system is a difficult one and required con­ siderable analysis of alternative methods. With more than 100 million vehicles on the road, it is of considerable importance to develop smooth, expeditious plans for a transi­ tion. It will obviously not be possible to have all vehicles on the road equipped im­ mediately. Therefore, . in the transition period during which vehicles are being equipped with polarized headlights, there will be both polarized and unpolarized headlights pres­ ent in traffic. Studies were made to determine what effect on visibility such mixed meetings would have (19). As can be seen in Figure 13, in only two instances would the resultant visibility con­ ditions in a meeting between a modified and unmodified vehicle be significantly worse than those presently encountered in a meeting of vehicles using conventional low beams (93 percent of all meetings occur with low beam). In both of these situations, the driver in the vehicle with the polarized equipment would be able to recognize quickly that, be­ cause he is seeing white lights (the sign of conventional unpolarized lamps), he is at a disadvantage. He could then simply move the analyzer out of his viewing field and he would revert to a conventional meeting situation with the same visibility obtainable at present, i.e., no worse than the present low-beam to low-beam meeting. Several conversion schemes for the transition are possible, and only by additional research and field testing will it be possible to develop the necessary information for wisely choosing among them. A scheme that is feasible, but may not be the optimum solution, is as follows: 1. During the first 3 years of the conversion, all new vehicles would be equipped with a polarized headlighting system. The conventional system would continue to be used for all meetings between vehicles that are not equipped, but where both vehicles

HEADLIGHT MQQ~S lJ DETECTION DISTANCE (FT) 100 200 300 400 500 600 700 OBSERVER ~ COMMENTS POSSIBLE TRANSITION S'ISTEM ,_ I I HIP LB +A FOR UNDER POWERED CARS ' ILB+A) I HIP +A HIP+A PROPOSED SYSTEM L_J _j_ _ OBSERVER MISUSING HB, HB+A HIP SHOULD SW ITCH TO LB I OBSERVER SHOULD REMOVE HIP+A LB+A ANALYZER ANO SWITCH TO LB, IF REQUESTED .I ,-, PRESEI\IT SYSTEM UP TO 7% HB HB Or MEEnNGS

OPPOSlf\lG VEHICLE COULJ L____ LB +A LBP SWITCH TO HIP AS OBSER1,ER HAS O.NALYZER __.r__ _ PRESENT SYS TEM 93% OF LB LB MEETINGS

POSSIBLE TRANSITION SYSTEM - -'- DETECTION TARGETS LB+A HIP FOR UNDER POWER t:O. RS (LB •A) = "NO PASSING" LINE OPPOSING VEHICLE SHOULD - J. HIP+A HB SWI TCH TO LB = PEDESTRIAN

OBSERVER SHOULD NOT USE 1-...--....J. -SIGN LBP+A LB ANALYZER, IE, LB TO LB

.!J HB - HIGH BEAM LB - LOW BEAM HIP- HIGH INTENSITY, POLARIZED +A - POLARIZATION ANALYZER IN USE

Figure 13. Mean target detection distances during transition period. 17 are equipped the polarized system would be used. Therefore, if one vehicle in a meet­ ing did not have the polarized system, the driver would flash his lights and the meeting would occur on low beam as at present. Unopposed driving would be with either high­ intensity polarized beams for new vehicles or high beam for used vehicles. 2. At the end of the third year, all drivers would be required to install analyzers in their vehicles. Used vehicles would be required to be equipped with the full system including the polarized lamps on title change. All new vehicles would continue to be equipped. Vehicles equipped with the polarized system would use it for meeting situa­ tions. All other vehicles would use low beam plus their analyzer. 3. At the end of the sixth year, conversion of all remaining vehicles would be required. This would mean that for the first 3 years of the program, owners of new vehicles would have the system on their vehicles, obtaining some limited benefit when meeting another new vehicle. By the end of the third year, approximately one-third of all ve­ hicles would be equipped (20). Because newer vehicles tend to be used more than older vehicles, equipped vehicleswould account for perhaps 45 percent of all oncoming traf­ fic. By fitting all vehicles with the analyzer at the end of the third year, it would in­ crease the benefit to those with fully equipped vehicles and would hasten voluntary con­ versions. By requiring retrofitting of used vehicles on resale for the next 3 years, conversion of the remaining vehicles would be speeded up, and by the end of the sixth year more than three-quarters of all vehicles would be fully equipped. These vehicles would account for 85 to 90 percent of all meetings. It is estimated that new vehicles would be equipped at an added cost to the owner of approximately $30 (19). To retrofit used vehicles would cost between $26 and $45 per vehicle depending onthe type of headlighting system already on the vehicle. Modifica­ tion of existing vehicles would be possible in most cases by any driver who is capable of making simple repairs such as adjusting headlights or changing spark plugs. A few vehicles probably could not be converted at the costs stated because of their low electrical power capacity. These vehicles could be equipped with an analyzer only and allowed to operate on low beam. Visibility for these drivers would be approxi­ mately the same as they now have (Fig. 13).

CONCLUSIONS AND RECOMMENDATIONS The research described in this paper has demonstrated that: 1. Polarization appears to be the only practicable method by which adequate vehicle headlighting can be provided without causing disturbing levels of glare to motorists going in the opposite direction; 2. Polarization is technically and economically feasible on today's vehicle population; 3. Polarization is advantageous in terms of improved visibility; and 4. The result of polarization will be enhanced vehicular control, safety, and com­ fort and probably improved traffic flow and utilization of highways at night. Although it has been generally concluded that polarization of headlights provides the most practicable approach to achieving adequate visibility at night, there are many as­ pects of the conversion and use of polarized headlights by the general public that re­ quire further consideration and evaluation. Therefore, it is suggested that a public test and evaluation program be undertaken to examine these problems in the real world and to develop precise data on the costs and benefits of such a system. The data to be obtained in such a test would evaluate public response to the system through interviews and operational studies of traffic flow and accident characteristics. Also, problems of equipment operation and maintainability would be studied. This public test should be conducted in isolation from traffic equipped with conventional headlights for a sufficient period of time to obtain reliable measures. An island with access from the mainland limited to car ferry, ocean-going shipping, and air transport would provide such re­ quired isolation ~). Such a test is currently being planned. 18

REFERENCES 1. Walker, W. P. Effects of Highway Lighting on Driving Behavior. HRB Proc., Vol. 20, 1940, pp. 511-521. 2. Normann, 0. K. Studies of Motor Vehicle Operation on Lighted and Unlighted Rural Highways in New Jersey. HRB Proc., Vol. 24, 1944, pp. 513-534. 3. Taragin, A., and Rudy, Burton M. Traffic Operations as Related to Highway Illu­ mination and Delineation. HRB Bull. 255, 1960, pp. 1-29. 4. Powers, Lawrence D., and Solomon, David. Headlight Glare and Median Width­ Three Exploratory Studies. Public Roads, Vol. 33, No. 7, 1965, pp. 125-142. 5. Schwab, Richard N. Night Visibility for Opposing Drivers With High and Low Head­ light Beams. Illuminating Engineering, Vol. 60, No. 5, 1965, pp. 364-372. 6. Cassel, Arno, and Janoff, Michael S. Identification and Evaluation of Remedial Aid Systems for Passing Maneuvers on Two-Lane Rural Roads, Vol. II. Tech. Rept. 1-201, Franklin Institute Research Laboratories, Feb. 1970. 7. Hare, Charles T., and Hemion, Roger H. Headlamp Beam Usage on U.S. Highways. Report on Phase III of a Study for U. S. Bureau of Public Roads by Southwest Research Institute, Rept. AR-666, Dec. 1968. 8. Hemion, Roger H. The Effect of Headlight Glare on Vehicle Control and Detection of Highway Vision Targets. Report on Phase I of a Study for U.S. Bureau of Public Roads by Southwest Research Institute, Rept. AR-640, May 1968. 9. Land, Edwin H., Hunt, J. H., and Roper, Val J. The Polarized Headlight System. HRB Bull. 11, 1948. 10. Johansson, Gunnar, and Rumar, Kare. A New System With Polarized Headlights. Department of Psychology, University of Uppsala, Sweden, Rept. No. 64, Dec. 1968. 11. Webster, L. A., and Yeatman, F. R. An Investigation of Headlight Glare as Related to Lateral Separation of Vehicles. University of Illinois, Eng. Expt. Sta. Bull. 496, 1968. 12. Road Research Programme, Cooperative Research Programmes. Research Group S2 on Lighting, Visibility and Accidents, Organisation for Economic Co-operation and Development, Doc. No. RR/S2/70.1, Jan. 1970. 13. A Policy for Geometric Design of Rural Highways, 1965. American Association of State Highway Officials, 1967. 14. Adams, Walter S. Effect of Vehicular Roll on a Polarized Headlighting System. In press. 15. Schiflett, Samuel G., Cadena, David G., and Hemion, Roger H. Driver Fatigue in Night Operation. Report on Phase II of a Study for U.S. Bureau of Public Roads by Southwest Research Institute, Sept. 1969. 16. Tsongos, Nicholas G., and Schwab, Richard N. Driver Judgments as Influenced by Vehicular Lighting at Intersections. Public Roads, Vol. 36, No. 1, April 1970. 17. Cadena, David G., and Hemion, Roger H. Disability Glare Effects From Vehicle Headlights and Fixed Overhead Street Lighting. Report on Phase IVB of a Study for U.S. Bureau of Public Roads by Southwest Research Institute, Rept. AR-700. 18. Hemion, Roger H. A Preliminary Cost-Benefit Study of Headlight Glare Reduction. Report on Phase V of a Study for U.S. Bureau of Public Roads by Southwest Re­ search Institute, Rept. AR-683, March 1969. 19. Hemion, Roger H. Disability Glare Effects During a Transition to Polarized Ve­ hicle Headlights. Report on Phase IV A of a Study for U. S. Bureau of Public Roads by Southwest Research Institute, Rept. AR-672, Jan. 1969. 20. Rear Lighting System Changeover Study, Volume II. Study for National Highway Safety Bureau by Systems Associates, Inc., NHSB No. HS-800011, May 1968. 21. Hemion, Roger H. Preliminary Site Selection for Public Test of Polarized Head­ lighting. Report on Phase VI of a Study for U.S. Bureau of Public Roads by Southwest Research Institute, Rept. AR-692. 19 DISCUSSION Merrill J. Allen, Division of Optometry, Indiana University The level of excellence and comprehensiveness of this paper is unusual. Schwab and Hemion conclude that the present headlight system of high and low beams is both inadequate (approaching 98 percent under some road conditions) and improperly used on 70 percent or more of the automobiles sampled on the highway. Fixed street light­ ing is noted to increase visibility beyond the power of the best mobile system foreseen in the paper. The authors compare a system of polarization of windshields and head­ lights with the current automobile headlight system and find that there are many bene­ fits and few disadvantages to the polarization system. One hazardous disadvantage is the reduction in brightness of oncoming headlights that causes them to be judged as farther away than they actually are. Lack of awareness of the importance of the bright­ ness clue to distance is evidenced by their statement: " ... care should be taken in the design of the analyzer to protect drivers during encounters from the side." We have observed that "protective" filters on the windshield or the spectacles may induce a person to venture onto a cross street or highway in front of cars that are too near and moving too fast to allow safe entry in front of them. The .authors did not explore the possible advantages of horizontal and vertical polar­ izer and analyzer planes rather than the usual 45-deg angle portrayed for such sys­ tems. Tilted windshields partially polarize the light in the 90-180 deg meridian. Pave­ ments have properties that may favor one plane of polarization over another for visi­ bility and protection from reflected glare. Rain drops, snow, or ice crystals in the air may dictate that windshield and headlight analyzers in a given vehicle be at right angles to one another rather than parallel as proposed. The polarization system chosen should take these factors into account. The authors evaluated the efficiency of beaded highway paint in their polarization system. The cube corner reflector and the beaded retroreflector over an aluminum substrate should also be evaluated. Because they must be used for signs and hazard marking and to protect pedestrians, retroreflectors must not be disadvantaged, for example, by turning off headlights and relying on street lighting as the authors have considered. Headlights, polarized or otherwise, must be retained regardless of street lighting-even including full daytime street lighting where headlights serve as running lights with a proven ability to reduce accidents (22, 23). The authors describe the presence of windshield dirt as being of little consequence, but it is not clear whether both inside and outside dirt was considered. Also, no mention is made of the age or condition of the surface of the windshield used in their studies. Service-station attendants and windshield wipers scratch the surface, and high-velocity airborne particles pit the surface. These defects must be evaluated because windshield damage appears to be serious enough to warrant changing the windshield every 30,000 to 50,000 miles. When using high-intensity sealed lamps equipped with polarizers and analyzers, speeds of 63 mph on dry pavement and 50 mph on wet pavement seem safe compared to speeds of 40 and 30 mph when using conventional headlights. Thus the proposed system would bring headlight range into agreement with the average speeds already found on our highways with no margin of safety. The Department of Transportation is considering a 95-mph speed governor on all motor vehicles. To achieve an appropriate visibility for 95 mph would require an incredible 16-fold increase in illumination over the proposed system! One solution to this dilemma not explored by the authors is the use of auxiliary side lights (24). This system offers the advantages of fixed street lighting on every highway and roadway since it casts light backwards into the opposing lane for use of oncoming drivers. Low-positioned lights on both sides of the vehicle reveal the details of the roadway for the oncoming driver and provide comfortable and sleep-fighting surround lighting for the vehicle itself. Because fixed street lighting permits target detection distances of about 4 times that with standard headlighting, the advantages of a vehicle­ borne system resembling street lighting must be considered for use in conjunction with the polarizing system. Otherwise, as highway speeds increase, the polarizing system by itself will be inadequate even before it becomes nationally operative. 20

Thus it appears to me that the limitation of a.11.y headlight system, polarized or not; becomes one of available power as highway speeds increase. A twofold increase in headlight range requires about 4 times the power and works welL only in a clean atmo­ sphere. Doubling highway speeds, however, requires about 4 times the stopping dis­ tance; Thus improvement of the current headlight design must be accompanied by more sophistication then polarizing alone can offer. It appears to me, however, that polari­ zation is a desirable and practical component of future automobile lighting designs. Certainly it is a significant improvement over the grossly inadequate system we now have, as so capably demonstrated by Schwab and Hemion.

References 22. Allen, Merrill J. Vision and Highway Safety. Chilton, 1970. 23. Allen, Merrill J. Seeing and Safety. Analogy, Allstate Insurance Co., No. 9, Fall 1970, p. 13. 24. Allen, Merrill J. Special Motor Vehicle Auxiliary Lights Designed to Reduce Effect fects of Glare and Aid Highway Visibility. Highway Research News, No. 5, May 1963, p. 24.

R. W. Oyler, Guide Lamp Division, General Motors Corporation I should be remiss, indeed, if I did not commend Schwab and Hemion for a thoughtful and well-presented paper. Perhaps it may give direction to more thorough assess­ ment of many of the claims and counterclaims that have been made about polarized headlighting in the last 50 years. The purpose of the remarks to follow is threefold: 1. To look at a few items on the check list to be used in further polarized head­ lighting investigations; 2. A.8 a ::,pecii'ic uui..~.ruw U1 uI otc:1.t~iut::iit 1, tu ougg~st that «. ~cc·v·u.lu.u.tiv~ vf the ::::c::;t factors be made both for original equipment and for retrofitting; and 3. To suggest an approach that might be considered for use in the near-term future while the longer r ange studies of polarized lighting sugge sted by the aut hors are being conducted. I particularly commend the authors for their recognition of the often-overlooked fact that no automotive lighting investigation is complete that does not include substantial evaluation under actual driving conditions on public highways. It is so easy for any of us to draw unfortunate conclusions if our work is confined exclusively to laboratories, abandoned air strips, or unopened portions of Interstate routes that the taxpayers have thoughtfully provided for us. Over the years, a number of questions have persistently recurred about the conse­ quences of polarized headlighting use; the extended test suggested by the authors should serve as an excellent opportunity to establish the validity of these questions or to lay them to rest for all time. Among these questions are the following: 1. How important is the loss of the halo or corona now put out by the lights of ap­ proaching cars on the far side of an intervening rise? 2. To what extent will "safe to pass" decisions be affected on two-lane roads? At present we are accustomed to looking at given light intensities. With the polarized ap­ proach we would be viewing a markedly decreased apparent intensity, and therefore we should find out what problems, if any, are generated in depth perception. 3. To what extent does snow or fog cause reflected light to be depolarized? If the degree is significant, the amount of light reaching the driver's eye through the analyzer could be further reduced. 21

4" The benefits of silhouette seeing will largely be removed. Is this a real loss? 5. In an overtaking situation, light comparable to that presently associated with upper beams would be going through the forward cars' backlights and impinging on their outside rear-view mirrors. Must the overtaking car dim? To do so negates much of the benefit of the new system; not to do so produces the condition that is now illegal for supposedly valid reasons. 6. In situations involving meeting and with cars also going in the same direction, will the reduced apparent intensity of rear presence and signal lights on preceding cars cause a dangerous situation? 7. What is the life expectancy of polarizers and analyzers? They are subject to varying degrees of attack from weathering, heat, and abrasion. Is their probable life at least equal to that of the vehicle? If not, how may their efficiency be monitored in motor-vehicle inspection programs? 8. Will city beam lamps be lighted at all times to guard against possible icing in inclement weather? If so, what should be their light distribution? Because cost-benefit relationships should join engineering and human factors on our evaluation list, it is appropriate to look at indicated cost differentials applying to original equipment and retrofitting of vehicles to install polarized headlighting. There is some reason to believe that the figures cited by the authors may well be conservative, although there are several ways that the original or retrofitting job could be done. In assigning any numbers to these equipment estimates and for purposes of these remarks, we will make certain assumptions. Among these are the following: 1. Cars would be equipped with four headlights to provide a driving beam or a city beam as needed. 2. About 2 ½ to 3 times as much light must originally be produced as at present in order to get "break-even" conditions for country driving. 3. To produce this amount of light, it will be necessary to provide an additional 100 to 150 watts for the driving beam. 4. Generators, switch gear, wiring, windshields, headlights themselves (including possible addition of 2 in some cases), polarizing media, and analyzers may have to be modified or additionally supplied. In the retrofitting mode, assessment must also be made of physical space availability for some of these. Subject to the foregoing and to the fact that cars have a wide range of power-demanding equipment, recent appraisal indicates that the cost estimates made in the major 1947 study of polarized lighting may still have validity. These were $ 30 to $ 80 for new cars and $100 to $ 200 for retrofitting. Even if we decrease these figures somewhat to reflect changes of the last 25 years, recognition that about $1.75 in 1971 is equivalent to $1.00 in 1947 still leaves us with an impressive package from a cost standpoint. Is there something that may be done now, so to speak, to provide a headlighting bene­ fit until such time as the studies recommended by the authors establish the parameters for the system we really want and should have? I suggest that an intermediate beam system, which can be used on both divided highways and two-lane roads, should receive consideration. In-house testing reveals the practicability of this approach and the di­ mensions of significant rather than incremental benefits that accrue from such an in­ termediate beam as compared to the present lower beam. In addition, the use of an upper beam with up to 150,000 cd should be favorably considered. After driving a car so equipped for over a year, I find that actual on-the-road experience has calmed many of my own fears about the consequences of this increased output. Although dimming is required at a greater distance, this is not all loss since silhouette remains. I have found, further, that the greater intensity tends to be self-policing in that instant respect is produced by my flashing an oncoming driver who had not previously considered it important to dim. In those situations when upper beam can properly be used, the 150,000 cd definitely add to driving safety and pleasure. I appreciate the opportunity to have discussed aspects of this paper and look forward to the results of further testing of polarizing lighting devices. 22 AUTHORS' CLOSURE. We would like to express our appreciation to the discussers and to thank them for the time spent in reviewing our paper. We were particularly gratified that both dis­ cussers agreed that the time has come to subject a polarized headlighting system to a controlled field test under driving conditions on public highways. The reservations of the discussers are generally predicated on an assumption that the analyzer is permanently in place on the windshield or always in use when driving at night. This is not the case. It is neither necessary nor desirable to look through the analyzer at all times. The driver should look through the analyzer only when op­ posing polarized headlights are in view and near enough to be disturbing, noticeable, or causing discomfort or visual degradation. Experience in our field research studies indicates that drivers quickly adapt to this concept of operation. Use of the analyzer in this manner negates all of the objections raised about loss of "over-the-hill" beam visibility of an approaching vehicle, inadequate or unfamiliar dis­ tance judgment of approaching headlights (or rear lights in following situations), loss of silhouette seeing, visibility in fog, and similar situations. It must be recognized that the analyzer's only function in the system is to reduce the glare caused by direct rays from polarized headlights that would otherwise be projected into the eyes of the observer. In response to Dr. Allen's suggestion that horizontal-vertical polarizers are better than the 45-deg system proposed, studies (8) show that comparative detection distances for a horizontal-vertical system are less th an for a 45-deg system. Other investiga­ tors have studied circular and elliptical polarization systems and these have been shown to be inferior to a 45-deg system. It was a conclusion of the first International Sym­ posium on Polarized Headlighting (Stockholm, 1969) that the 45-deg system provided the best approach, and future development should be confined to this system. Some minor deviation from this configuration may be necessary to achieve optimum perfor­ mance in view of the effects of curved and slanted windshields, and further work will be needed in this area. With respect to comments by Mr. Oyler, there is not agreement as yet with respect to the necessity for retaining unpolarized low-beam headlights for urban (lighted road- ,. __ ,,,_.,.\ ,..1_..· ..... • ..... ~ ,,-,1...,.;,... .: ...... -+.: ...... 1,..,....,. .,....,.,.,...,...,...... +,....,..,. ....,.....,.,....1-,1 ...... ,...... :-1-1-. + ...... 1 ...... ,....,... nhn i::.a + ...... VVU.JJ U.L.LV.L.J..lE,• .J...l.L.1...:)' .LJ.J. }'U....Ll-.1.\..,UJ..U...L' ,1-'.&.\..,UV.lJ.l,U U.. fJ.LV,._.,.L\..,J.,lJ, VV.&.l,.L.l ._,VWV .LU...I..L.1.1-', ..L.1..a..J..l, VV l-J}''-- systems, If the analyzer is not used in urban driving, there would be no difference in appearance between the polarized and unpolarized low beams. If the driver desired to use his analyzer, visibility might be enhanced or reduced, if the vehicles he faced were equipped with polarized low-beam lights, depending on the level of fixed lighting present. There is also some thinking at present of the use of a dim-dipped low-beam system for use on well-lighted urban roadways. It would seem that polarized low beams might provide even better visibility than this approach. Dr. Allen's analysis of sixteen times present polarized lighting power requirements for safe visibility at 9 5 mph is not considered valid. The AASHO formula stopping distance for the 85th percentile driver and vehicle is 1,020 ft on dry pavement. (No­ body in his right mind would drive at 95 mph on a wet pavement.) To achieve this visi­ bility distance with the polarized system in a meeting situation that now provides a "safe" speed of 6 5 mph (460 ft) would require (1, 020/ 460) 2 or 4. 7 times the maximum beam intensity of 65,000 cd used in these studies. The suggested system of vehicle side lighting proposed by Dr. Allen would provide no benefit for unopposed driving and is not comparable to fixed roadway lighting in such cases. Further, it could contribute to visibility in the meeting situation only if the headlights were coincidentally reduced in intensity to decrease veiling glare. The problem of glare from polarized high-beam headlights of a following vehicle, it would appear, can be resolved by the use of an analyzer applied directly on the face of the rear-view mirrors in the "extinction mode." This should provide improved visi­ bility over present types of prismatic night mirrors. Only the direct beams of the following headlights would be diminished while unpolarized surround lighting would still be transmitted at 25 to 30 percent efficiency rather than at 8 to 10 percent trans­ mission as is the case with present designs of night driving mirrors. 23

Mr. Oyler raises important questions of operating techniques, life, reliability, ser­ viceability, and maintenance that cannot be answered today. These are some of the primary objectives of the pr oposed public test of polar ized headlighting that, along with evaluation of expected operational benefits of enhanced safety, impr oved traffic flow, and increased comfort in night driving, make the proposed public lr ials so imperative as the next phase of this program.