Click for next page ( 23

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 22
22 CHAPTER 4 Performance Experiment From the results of the static screening experiment, four presented at the same distances without the truck as a baseline warning-light configurations were selected for further testing in measurement for pedestrian detection. dynamic conditions. Three measures considered the nighttime identifiability of a maintenance vehicle with warning lights; the Glare. The glare independent variable provided insight detectability of a pedestrian standing close to the maintenance into the ability of the warning-light system to be visible when vehicle; and the ranking of the vehicle lighting in terms of other (and particularly opposing) vehicles are present. The discomfort glare, attention-getting, and urgency. The study in- glare was simulated using the headlights of a parked vehicle cluded an uninformed trial where participants viewed the light- close to the experimental truck. ing systems without knowing the nature of the study and also Weather. The weather independent variable provided incorporated adverse weather in the testing conditions. insight on which type of warning light performs best in real- istic weather conditions. The three levels were dry, rain, and Experimental Methods fog. Rain and fog were kept consistent among participants by using a weather-making system to control the levels. Experimental Design Ambient Lighting. The ambient lighting independent The choice of independent variables was driven by the results variable provided insight on which type of warning light of the static screening experiment, and the need to provide performs best in daytime and nighttime conditions. realistic test scenarios that drivers are likely to encounter in everyday driving. The experiment included between-subjects variables (gender and age, each with two levels) and within- Assignment of Treatments subject variables (warning light and pedestrian, each with four levels; glare and ambient lighting, each with two levels; and Treatments were balanced across age and gender to elimi- weather with three levels). Because of the large number of vari- nate presentation bias among the groups. A balanced Latin ables, a mixed-factor partial factorial design was used to allow Square was used to create the orders of presentation for each exploration of the most relevant main effects and interactions. driving session. Eight orders were used for nighttime sessions, The final experiment design resulted in a total of 116 conditions with each participant within an age and gender group receiv- for each participant: 40 during the day, and 76 during the night. ing a different order. Four orders were used for daytime ses- sions, with two participants within an age and gender group receiving the same condition. Within-Subject Variables Warning Light. Four warning light configurations were Dependent Variables used: high-mounted rotating beacon, low-mounted rotating beacon, LED, and strobe. Seven dependent variables were used in this investigation. The dependent variable of lane-change distance was tested Pedestrian. The pedestrian was presented at two different only during the uninformed trial, which was at night in clear distances from the experimental truck to see how each light weather. The dependent variables of vehicle identification affected the visibility of the pedestrian. A pedestrian was also distance, pedestrian detection distance, and discomfort glare

OCR for page 22
23 rating were tested only during nighttime conditions (night dry, conditions when there was a low degree of contrast between night rain, and night fog). The dependent variables of attention- the lights and the background. getting rating and confidence rating were tested only dur- ing daytime conditions (day dry and day fog). The dependent Attention-Getting. To establish the effectiveness of the variable of urgency rating was tested during all nighttime warning lights, a metric for conspicuity was established. This conditions and during daytime fog conditions. was done using a seven-point rating scale ranging from "not at all attention getting" to "extremely attention getting." This Lane-Change Distance. During the uninformed trial, scale was administered during daytime conditions. participants were forced to pass the slow-moving dump truck (experimental vehicle), unaware that the truck was involved in the study. The distance at which the participant Participants initiated the lane change to pass the truck was marked as the Thirty-two subjects, 16 males and 16 females, were selected lane-change distance. to participate in this study. The subjects were evenly selected Vehicle Identification Distance. To establish each warn- from two age groups of 25 to 35 years of age and 65+ years ing light's ability to alert a driver to the presence of a mainte- of age. IRB approval was obtained prior to recruiting sub- nance vehicle, the distance at which a participant could identify jects. When subjects arrived for the first session, they signed the light source as belonging to a vehicle was recorded. The dis- an informed consent form before beginning any experimen- tance traveled between this point and when the participant tal activities. Subjects were paid $20/h for each driving session, vehicle passed the experimental truck was defined as the vehi- and a $30 bonus if they completed all four driving sessions. cle identification distance for the warning light on display. This They were allowed to withdraw at any point in time, with measurement was taken during each nighttime condition. compensation adjusted accordingly. Pedestrian Detection Distance. To establish each warn- ing light's ability to allow a driver to see maintenance workers Apparatus near a maintenance vehicle, the distance at which a participant Test Road could detect a pedestrian standing near the experimental dump truck was recorded. The pedestrian detection distance for the The experiment took place on the Smart Road--a 2.2-mile- warning light on display was defined as the distance traveled long controlled-access, two-lane road. A 0.5-mile section of from this point to the point at which the participant vehicle the Smart Road is equipped with an artificial weather-making passed the pedestrian. This measurement, along with a base- system that was used to create the rain and fog conditions for line measurement, was taken during each nighttime condition. this study. A degree of control was attained by not allowing pub- lic vehicles and pedestrians to enter the Smart Road and by con- Urgency. The urgency rating dependent variable mea- trolling the level of the rain and fog so that it was consistent sured the level of urgency that subjects felt was conveyed by among participants. the warning lights. A five-point Likert-type rating scale with end points "not at all urgent" and "totally urgent" was used to capture their ratings. This scale was administered during each Test Vehicles nighttime condition and the daytime fog condition. Like the static experiment, two test vehicles were used in Discomfort Glare. The discomfort glare rating dependent this experiment: a participant vehicle and an experimental variable measured the discomfort experienced by the subjects vehicle. The experimental vehicle was the same VDOT dump when presented with the warning lights. A nine-point rating truck that was used in the static experiment. However, the scale with end points "not noticeable" and "unbearable" was dump truck was outfitted with the four warning lights of used to capture their ratings. This scale was administered only interest (Figure 4). at nighttime when discomfort glare is at its worst due to the high The high beacon lights were placed above the cab of the degree of contrast between the lights and their backgrounds. truck, one on each side. The low beacons were placed on small shelves on the back of the tailgate, one on each side. The strobes Confidence. The confidence rating dependent variable and LEDs were mounted on a rack on the tailgate, one on measured the confidence level of the subjects that they could each side. see the warning light. The scale was a five-point Likert-type All lights were manually controlled by an operator who rating scale with end points "not at all confident" and "totally sat in the cab of the truck. There was radio communication confident." This scale was administered only during daytime between the participant vehicle and the experimental vehicle

OCR for page 22
24 Figure 4. VDOT dump truck outfitted with warning lights. that allowed the in-vehicle experimenter to prompt the light A baseline pedestrian was also presented several times for operator for the next light. The participants drove a 2002 each participant. This pedestrian would follow the same pro- Cadillac Escalade; an in-vehicle experimenter rode along to cedures, but stood on a section of road away from the dump provide directions and to record data. Vehicle identification truck. This allowed for a comparison of detection distances. distance and pedestrian detection distance were recorded using a data acquisition system (DAS) installed in the vehicle. Stimuli Attention-getting, discomfort glare, urgency, and confidence ratings were recorded by hand on an order sheet and later The stimuli used for this experiment were commercially entered into a spreadsheet. available light sources acquired from manufacturers. The The DAS recorded vehicle network data such as accelera- light sources were selected based on their performance in tion and speed, as well as four camera angles and information the static screening experiment, photometric characteristics, entered by the in-vehicle experimenter such as the partici- and the suitability for the experiment. pant number, participant age, experimental order, and button presses. High-Mounted Beacon. The high-mounted beacon used was a PSE Amber, model 550 FRAMH 12 V 100 W. A sum- mary of the rotating beacon light characteristics is provided in Pedestrian Appendix C1. The pedestrian was an on-road experimenter who wore Low-Mounted Beacon. The low-mounted beacon used denim surgical scrubs and a VDOT-issued reflective safety was also a PSE Amber, model 550 FRAMH 12 V 100 W, also vest. Depending on the presentation order, the pedestrian provided in Appendix C1. would either stand 40 ft or 80 ft behind the dump truck in the center of the lane for each lap driven by the participant. LED. The LED used was an amber Whelen 500 Linear When the participants verbally indicated that they could LED Flash Light. One light on each side of the truck was used. see the pedestrian, the in-vehicle experimenter would say The LEDs were displayed in a 1 Hz asynchronous pattern. "clear" over the radio. This was the pedestrian's signal to A summary of the LED light characteristics is provided in clear the road. If for any reason the in-vehicle experimen- Appendix C1. ter did not give the clear signal, the pedestrian would clear the roadway automatically when the vehicle reached a pre- Strobe. The strobe used was an amber Whelen 500 Linear determined proximity. Once the participant vehicle turned Strobe Light. One strobe light on each side of the truck was around and passed the dump truck on the way back to the top used. The strobes were displayed in a 1 Hz asynchronous pat- of the road, the pedestrian would get into position for the tern. A summary of the strobe light characteristics is provided next lap. in Appendix C1.

OCR for page 22
25 Methods vehicle present and one without. A second pedestrian was oc- casionally presented on a section of road not near the truck Upon arriving at VTTI for the first driving session, each in order to get a baseline pedestrian detection distance. participant was asked to read and sign the Information Sheet In addition, participants were also asked to rate the warning (Appendix E1), and fill out a health and vision screening ques- lights in terms of discomfort glare and urgency at two distances tionnaire. Each participant was also required to take an in- (2400 ft and 1200 ft). This procedure was also done twice: formal visual acuity test using a Snellen chart. The vision test once with a glare vehicle and once without. The first ratings was performed to ensure that all participants had at least were done after the first four laps, and the second ratings were 20/40 vision, which is the legal minimum to hold a driver's li- done after the last four laps. Depending on the presentation cense in Virginia. Participants were tested for color blindness order being used, a glare vehicle would either be present for using pseudo isochromatic plates, but were not excluded based the first two ratings or for the last two ratings. on results. The rain towers were then turned on, and the same steps were At the beginning of each driving session, the participants repeated for the nighttime rain condition. The participant's were given an information sheet that explained that they would speed limit was reduced from 35 mph to 25 mph for safety be expected to drive on Main Street in Blacksburg and on the because of the wet road surface and decreased visibility of the Smart Road under various weather conditions. It also outlined pedestrian. The distances at which the discomfort glare and the risks involved, and their rights and responsibilities as par- urgency ratings were recorded were also reduced (to 1200 ft ticipants. Before the participants' first driving session, they and 600 ft). Once all laps and ratings were complete, partici- would sign and date the information sheet. Upon arriving at pants were instructed to return to the VTTI building where VTTI for each subsequent driving session, the participants they were compensated for their participation. were asked to review the same sheet, and initial and date it for each visit. Second Driving Session: Daytime Dry First Driving Session: Surprise, Nighttime Dry, During the daytime sessions, only subjective ratings of the and Nighttime Rain warning lights were collected. Ratings were taken at four dis- tances (4800 ft, 3600 ft, 2400 ft, and 1200 ft) and in two direc- The first driving session consisted of three parts: the sur- tions (facing downhill and facing uphill). These directions were prise, nighttime dry, and nighttime rain trials. During the used because of the difference in contrast of the lights and their surprise trial, the participant was unaware that the focus of background. With the uphill view of the experimental truck, the study was on vehicle warning lights. Participants were the warning lights on top of the vehicle were visible against the instructed to drive on Main Street towards Blacksburg, where sky, whereas for the downhill view, the lights appeared against the experimental dump truck was waiting ahead. Instructions a foliage background as shown in Figure 5. read to the participant were designed to force them into For this session, the participants were first asked if they could passing the truck. The lane-change distance was defined as see the warning light being displayed. If they answered "yes," the distance between the participant vehicle and the truck at the they were asked to rate how confident they were that they saw moment a lane change was initiated. This trial was followed it using the confidence rating scale. Finally, the participants up by a questionnaire, which was administered in the nearest were asked to rate the attention-getting nature of the light using parking lot. The participant was then debriefed on the true the attention-getting rating scale. The participants would an- purpose of the research and signed an informed consent form swer these questions about each warning light at each distance. for continued participation. The downhill ratings were collected first, followed by the uphill During the nighttime dry trials, the participant drove on the ratings. Upon completion, the participants were instructed to Smart Road toward the experimental dump truck that was dis- return to the building where they were paid for their time. playing one of the four warning lights. The participant would verbally indicate when he or she could identify the light source Third Driving Session: Nighttime Fog as a vehicle, and when he or she could detect a pedestrian in the roadway near the truck. These points were marked by the During the nighttime fog session, the participants followed in-vehicle experimenter in the DAS data by use of a push the same protocol as the nighttime rain session. For safety, the button. The distance between the participant vehicle and the participants were instructed to drive at 15 mph while in the identified target (i.e., the dump truck or pedestrian) was fog. Each participant drove a total of eight laps on the Smart defined as the vehicle identification and pedestrian detection Road, indicating when they could identify the light source distances for the warning light on display. This procedure was as a vehicle and when they could detect the pedestrians in the repeated twice for each warning light: one trial with a glare road. After the fourth and eighth laps, the participants were