Pedestrian Visibility at Night: Effects of Pedestrian Clothing, Driver Age, and Headlamp Beam Setting
چکیده
Reduced visibility is believed to be a major contributor to night-time pedestrian fatalities. In this study, the ability of drivers to recognize the presence of roadside pedestrians was determined and recognition distances measured using a parallax-based device. Ten young and ten older participants drove at night around a closed road circuit and pressed a button whenever they recognized a pedestrian. Four pedestrian clothing conditions were tested (black, white, retroreflective vest and retroreflective material in the ‘biomotion’ configuration) in both low and high beam conditions. Two pedestrians walked in place on the shoulder of the road, one in darkness (primary) and one in a high glare position (secondary). Response distances were recorded for the primary pedestrian. Overall, drivers identified only 61% of pedestrians in the presence of glare and 76% in the absence of glare, with only 5% of darkly clad pedestrians being identified in the glare condition. Averaged across conditions, the primary pedestrian was identified at 76.5m (251ft), with distances being shortest for darkly clad pedestrians (13m (43ft)), and greatest for the ‘biomotion’ pedestrian condition, which improved response distances by up to 50X. The vest condition failed to produce similar benefits, despite having an equivalent amount of retroreflective material. Overall, older driver response distances were 58% of those of the younger drivers. These results demonstrate that while all drivers have significant difficulty recognizing pedestrians at night, elderly drivers experience even greater difficulty. The results also demonstrate the value of clothing configurations that enhance the perception of biological motion. TRB 2003 Annual Meeting CD-ROM Original paper submittal – not revised by author. Wood, Tyrrell & Carberry 3 INTRODUCTION Pedestrian fatalities are a major road safety issue. In the United States, 78,000 pedestrians were injured and 4,739 pedestrians were killed in the year 2000; pedestrian fatalities accounting for 85% of all non-occupant fatalities in that year (1). One of the basic driver errors responsible for such collisions is believed to be the late detection of other road users (2). Fatal pedestrian collisions are over-represented at night, where nearly two-thirds (64%) of all fatal pedestrian collisions occur under night-time conditions (1) and more than half of all pedestrian deaths and injuries occur when pedestrians cross or enter the road system (3). Reduced visibility is believed to be one of the major contributors to pedestrian collisions that occur at night (4). This is supported by recent studies which have also demonstrated that crashes involving pedestrians are most common in the dark time periods of daylight saving transitions, in some cases showing up to seven times more risk at night compared to daytime (5). In contrast, this study found that singlevehicle run-off-road crashes showed little difference between light and dark time periods, suggesting factors other than light level play the dominant role in these crashes. The conspicuity of pedestrians at night can be enhanced by visibility aids such as street lighting, reflectors, reflective clothing and lights. Retroreflective garments, such as vests and jackets are also widely used by construction workers, police and emergency personnel who have greater exposure at night-time, in order to enhance their visibility and safety. Retroreflective material positioned on the moveable joints to create the sense of “biological motion” has also been advocated as a means of enhancing the visibility and recognition of pedestrians (6). There have been a number of studies that have sought to determine how the use of such visibility aids might enhance pedestrian visibility. Many of these studies have been undertaken in the laboratory, using either slide or video-based representations of pedestrians wearing different clothing combinations. Although these provide good baseline data on which to design field studies, they do not necessarily represent the lighting or environmental conditions, vehicle or pedestrian motion present in real road conditions, nor do they necessarily provide a good sense of face validity. A series of studies have been undertaken under either closed or open road situations where the participants acted as passengers (7,8). Other studies have used participants as drivers on closed road circuits, however, many have used distance measurement techniques which have potential limitations (9), or have used driving populations made up entirely of undergraduate or members of institutions who are not necessarily representative of the general driving population (10). In the study reported here, we determined the percentage of drivers who recognized pedestrians walking in place on the shoulder of a real road, in the absence (primary pedestrian) and presence of glare (secondary pedestrian). The recognition distances for the primary pedestrian were systematically recorded using a novel parallax-based measurement device. Participants included people recruited from the wider community of a range of different ages. The effect of different pedestrian clothing conditions and headlight beam were also investigated systematically. METHODS Participants There were a total of 20 participants divided into two groups of ten drivers (mean ages: 27.8 ± 4.7 yrs (range 21-34 yrs), and 67.9± 5.30yrs (range 60-75 yrs)). All participants were licensed drivers with at least three years of driving experience, and all reported that they drove regularly. All participants passed the minimum drivers’ licensing criteria for binocular visual acuity of 6/12 (20/40). Experimental Vehicle & Closed Road Circuit The experiment was conducted under night-time conditions on a closed road circuit which has been used in previous studies of driving and vision (11). The circuit consists of a bitumen road, with hills, curves, bends and straight sections and standard road signs and road markings. The circuit does not include any street lighting; a 1.8 km (1.1 mile) section of the circuit was used for this study. A series of glare lights, consisting of a set of stationary battery-powered headlights were positioned at three locations along the circuit to simulate the effect of oncoming headlights. These headlights were activated when the test vehicle drove through a remote sensor. A series of 50 cones with retroreflective markers distributed along the circuit acted as distracters and provided a degree of visual clutter. The experimental vehicle was a 1997 Nissan Maxima, which had been serviced (including headlamp alignment) immediately prior to the experiment. Two synchronized digital video cameras were mounted a fixed distance apart on the roof of the vehicle. This system produced two overlapping digital images of the road scene, and was linked to a fibreoptic marking system and response button which recorded the exact moment that the participant recognised a pedestrian, thus identifying the relevant stereoscopic pairs of images. These images were analysed off-line to determine the positions of a corresponding point in the marked images. The change in position of these points is the parallax, which was TRB 2003 Annual Meeting CD-ROM Original paper submittal – not revised by author. Wood, Tyrrell & Carberry 4 measured and the recognition distance thus calculated. The technique has a high level of accuracy and validity, the measured accuracy of the system being better than 5-10% of the true distance (12,13). To minimize measurement error further, traffic cones with retroreflective markers had been positioned strategically in advance. Thus when a participant driver pressed the button to indicate that a pedestrian was recognized, the measurement system was only required to measure the distance from the test vehicle to the nearest traffic cone. The distance from that cone to the pedestrian was known in advance and was added to the measurement. Four experimenters were involved in any one experimental run. Two experimenters were seated in the experimental vehicle, while two experimenters acted as pedestrians located at different positions on the circuit. Both pedestrians were positioned on one shoulder of the roadway; the test vehicle was driven towards the pedestrians in the far lane, such that there was at least one lane of travel separating the pedestrian and the test vehicle. One pedestrian was situated at the end of a 400 m (1312 ft) straight section of the track and was referred to as the primary pedestrian. The secondary pedestrian stood on an opposite portion of the circuit, 10.2 m (33.5 ft) behind one of the three glare lights. Each pedestrian had a two-way radio, as did the backseat experimenter. All communication was conducted between trials with the experimenter outside of the vehicle, so the subject could not hear the conversations. Clothing and headlight beam conditions For each experimental run the pedestrians wore one of four clothing conditions, black, white, vest or biomotion: • Black: a black cotton sweatshirt (2% reflectance), a pair of black cotton sweatpants, black gloves and black shoe covers. • White: a large white cotton lab coat (68% reflectance) with white gloves and white cotton leggings. • Vest: the clothing from the Black condition described above with the addition of a retroreflective panel measuring 30cm x 17.5cm (525 cm) worn on the chest. • Biomotion: the clothing from the Black condition with the addition of retroreflective straps (2.5 cm (1 inch)) around the wrists, elbows, shoulders, waist, knees and ankles. The total area of visible retroreflective material was matched to the Vest condition (525 cm). Each pedestrian wore each of the four clothing conditions twice, once for a low beam lap and once for a high beam lap. Pedestrian clothing and headlamp beam setting were randomized. The primary pedestrian was always absent on the third lap and the secondary pedestrian was always absent on the seventh lap. Pedestrians walked in place as the test vehicle approached. This allowed the inclusion of natural pedestrian motion while keeping the pedestrians in a known location. Procedures Each participant completed ten laps of the test circuit. The first lap was a practice lap; all data were collected on laps 2-10. To increase driver workload, participants were instructed to read aloud all road signs that they encountered. Their only other instructions were to follow the prescribed route, to drive at a comfortable speed, and to press a large (6 cm x 12 cm) luminous dash-mounted response button as quickly as possible each time they recognized that a pedestrian was present. They were instructed not to press the button until they were confident that what they saw was in fact a pedestrian. They were also informed that there wouldn’t always be pedestrians on the circuit. Two primary dependent variables are reported here. First, the percentage of trials in which pedestrian recognition occurred is reported. Pedestrian recognition was recorded as having occurred if the driver pressed the response button at any point along the approach to a pedestrian or immediately after having passed a pedestrian. This procedure is conservative in that recognition here does not imply that the driver would have been able to initiate a successful avoidance maneuver. Recognition was measured for both the primary pedestrian (no glare) and for the secondary pedestrian (glare). The second dependent variable is the driver’s response distance, which is defined here as the distance from the vehicle to the primary pedestrian (no glare) at the moment the response button was pressed. Response distances were coded as zero for all trials in which the driver did not recognize the pedestrian or had passed the pedestrian before pressing the button. Detection distances (the driver’s response distance plus the distance traveled during the driver’s reaction time) were also recorded but are not presented here. Response distances to the secondary pedestrian were not measured. RESULTS Tables 1 and 2 present the percentage of drivers who correctly identified the presence of the pedestrian, as a function of pedestrian clothing, the beam condition and the age of the driver. The data are given for the identification of pedestrians in the absence of glare (primary pedestrian, see Table 1) and in TRB 2003 Annual Meeting CD-ROM Original paper submittal – not revised by author. Wood, Tyrrell & Carberry 5 the presence of glare (secondary pedestrian, see Table 2) and demonstrate that overall, drivers correctly identified only 61% of pedestrians when glare was present, while 76% were identified in the absence of glare. Importantly, when the data were considered in terms of driver age, less than half of the pedestrians were identified in the presence of glare by the older drivers (48%) compared to 75% for the younger drivers. The data also clearly demonstrate that the darkly clad pedestrians were identified least often, regardless of condition or driver, with only one driver (5%) recognizing the pedestrian wearing black in the presence of glare when the headlamps were on low beam. Interestingly, recognition in these conditions increased to 35% when high beams were used. Also interesting is the fact that recognition of the pedestrians wearing white clothing was equal to or better than recognition of the pedestrian wearing the retroreflective vest. When the same amount of retroreflective material was worn in the biomotion configuration, however, the pedestrians were identified by 100% of the young drivers, and by 75% of the older drivers, even when the glare challenge was present. The percentage of pedestrians seen was modeled as a function of pedestrian clothing, headlamp beam, presence or absence of glare and driver age using a logistic regression model. This indicated that pedestrian clothing (χ3 = 64.5, p 8 REFERENCES 1. National Highway Traffic Safety Administration. (2002). Traffic safety facts 2000: Pedestrians. Washington, D.C.: U.S. Department of Transportation. http://www.nhtsa.dot.gov/people/ncsa/pdf/newPed99.pdf Accessed July 2nd, 2002. 2. Rumar K. (1990) The basic driver error: late detection. Ergonomics, 32, 1281-1290 3. National Safety Council Accident Facts. Itasca, IL: 1994. 4. Owens, D.A., & Sivak, M. (1996). Differentiation of visibility and alcohol as contributors to twilight road fatalities. Human Factors, 38, 680-689. 5. Sullivan, J.M., and Flannagan, M.J. (2002). The role of ambient light level in fatal crashes: inferences from daylight saving time transitions. Accident Analysis and Prevention, 34, 487-498. 6. Owens, D.A., Antonoff, R., & Francis E.L (1994). Biological motion and nighttime pedestrian conspicuity. Human Factors, 36, 718-732. 7. Allen, M.J., Hazlett, R.D., Tacker, H.L., & Graham, B.V. (1968, Oct.). Actual pedestrian visibility and the pedestrian’s estimate of his own visibility. Proceedings of the Thirteenth Annual Conference of the American Association for Automotive Medicine, 293-299. 8. Luoma, J., Schumann, J., & Traube, E.C. (1996). Effects of retroreflector positioning on nighttime recognition of pedestrians. Accident Analysis and Prevention, 28, 377-383. 9. Chrysler ST, Danielson SM, Kirby VM. (1996) Age differences in visual abilities in nighttime driving field conditions. Proc Human Factors Ergonomics Soc 40th Meeting, 923-927. 10. Blomberg, R.D., Hale, A., & Preusser, D.F. (1986). Experimental evaluation of alternative conspicuity-enhancement techniques for pedestrians and bicyclists. Journal of Safety Research, 17, 1- 12. 11. Wood, J.M. & Troutbeck, R. (1994). The effect of visual impairment on driving. Human Factors, 36, 476-487. 12. Jones, K., Bentley, B.E., Wood, J.M. & Woolf, M.I. (1998). Application of parallax for the measurement of visibility distances in the open-road environment. International Archives of Photogrammetry and Remote Sensing. Vol XXXII. Part 5. 74-79. 13. Jones, K., Wood, J.M., Woolf, M.I. & Bentley, B.E. A technique for on-road assessment of road sign visibility distances. Vision in Vehicles VII. In press. 14. Tyrrell, R.A. & Patton, C.W. (2000, January). Using education to influence pedestrians’ estimates of their own nighttime visibility. Presented at the 79 Annual Meeting of the Transportation Research Board, Washington, D.C. 15. Brooks, J.O., Tyrrell, R.A. & Olson, H.M. (2001, January). Pedestrian Visibility at Night: Evaluating a Method of Teaching Young Drivers to be Safer Pedestrians. Presented at the 80 Annual Meeting of the Transportation Research Board, Washington, D.C. 16. Owens, D.A., Wood, J.M., Whittam, D., Woolf, M. Effects of age and illumination on driving performance. Vision in Vehicles VIII. In press. 17. Mortimer, R.G., & Olson, P.L. (1974). Evaluation of meeting beams by field tests and computer simulation. Report No. UM-HSRI-HF-74-27, Ann Arbor, MI: The Highway Safety Research Institute. 18. Shinar, D. (1984). Actual versus estimated nighttime pedestrian visibility. Ergonomics, 27, 863-871. 19. Leibowitz, H.W., Owens, D.A., & Tyrrell, R.A. (1998). The Assured Clear Distance Ahead Rule: Implications for traffic safety and the law. Accident Analysis and Prevention, 30, 93-99. 20. Gish, K.W., Shoulson, M., Perel, M. (2001, January) Driver behavior and performance using an infrared night vision enhancement system. Presented at the 80 Annual Meeting of the Transportation Research Board, Washington, D.C. TRB 2003 Annual Meeting CD-ROM Original paper submittal – not revised by author. Wood, Tyrrell & Carberry 9