|
Effects of Partial and Total Sleep Deprivation on Driving Performance
by Robert D. Peters,
Esther Wagner, Elizabeth Alicandri, Jean E. Fox, Maria L. Thomas, David
R. Thorne, Helen C. Sing, and Sharon M. Balwinski
Background The National Sleep Foundation estimates that more than two-thirds of American adults have a sleep-related problem and that 23 percent have actually fallen asleep while driving. These shocking statistics are based on data collected in a national telephone survey of 1,027 Americans in late 1997 and early 1998. To compound the basic problem of sleep deprivation, 86 percent of the participants in the survey failed a basic "sleep IQ test" that asked questions such as "Does boredom make you sleepy?" and "Does the human body adjust to night-shift work?" Many sleep-deprived people erroneously believe that they can function well with only a few hours of sleep per night over a lengthy period. But sleep deprivation, even partial sleep deprivation, has significant negative effects on mental and physical performance, including driving. The National Highway Traffic Safety Administration estimates that from 1989 through 1993, driver drowsiness/fatigue was a contributing factor in 100,000 crashes annually on U.S. highways.2 During the same five-year period, drowsiness/fatigue was cited as a factor in an annual average of 1,357 fatal crashes (3.6 percent of all fatal crashes). As a result, about 1,544 people were killed each year.3 These statistics for driver sleepiness and accidents are considered conservative because of differences in state reporting practices, lack of firm evidence about the cause of many crashes, and failure of drivers involved in crashes to report that sleepiness played a role. The Federal Highway Administration's (FHWA) Office of Motor Carriers (OMC) has made driver drowsiness/fatigue the dominant human factors research issue in its research and technology (R&T) program. OMC has more than 25 completed, ongoing, or planned R&T projects related to driver drowsiness/fatigue and hours-of-service regulations. Driver fatigue is a safety issue of special concern to the commercial motor vehicle (CMV) transportation community. CMV drivers may drive up to 10 hours continuously before taking a break, often drive at night, and sometimes have irregular and unpredictable work schedules. Much of their mileage is accumulated during long trips on interstate and other four-lane roadways. Because of their extensive mileage exposure and other factors, commercial drivers' risk of being involved in a fatigue-related crash is far greater than that of non-commercial drivers C even though CMV drivers represent only about 4 percent of the drivers involved in known fatigue-related crashes and rate of involvement per mile traveled is no greater than that of non-commercial drivers. Driver drowsiness/fatigue has also been recognized as a problem for military drivers. According to the U.S. Army Safety Center, approximately 9 percent of the wheeled-vehicle crashes that resulted in injury or death during Operation Desert Shield and Operation Desert Storm were attributed to driver drowsiness/fatigue.4 Reducing the extent of the drowsy driver problem is critical to improving the safety of our nation's highways. To address this issue, the research community must determine which driving performance measures are sensitive to identifying sleep deprivation in vehicle operations. These findings will be used to develop a method of predicting when sleepiness can put drivers at risk for crashes. Once these metrics are developed and refined, countermeasures to alleviate the drowsy driver problem can be generated and tested.
A study to investigate the effects of sleep deprivation on driving performance was conducted jointly by FHWA's Human Factors Laboratory and the Walter Reed Army Institute of Research (WRAIR). The study examined the effects of progressive sleep deprivation on driving performance to assess the rate of crashes and the changes in driving performance resulting from sleepiness. Because it would be unsafe to study this under real driving conditions, the high-fidelity highway driving simulator (HYSIM) at FHWA's Turner-Fairbank Highway Research Center (TFHRC) in McLean, Va., was used. A variety of measures, including continuous electroencephalogram (EEG) monitoring, videotaping, and analyses of driving performance data and questionnaire data were used to determine the effects of sleep deprivation on driving performance. Method Subjects remained in a residential suite with testing chambers at WRAIR during the eight-day study and made a total of five trips to TFHRC for training and tests in the HYSIM. On the first day, subjects drove the HYSIM for training purposes only. On the second through fifth days, data were collected. Table 1 provides a description of the daily test conditions. Driving Simulator and Test
Scenario On each of the four test days, subjects entered the HYSIM scenario at a different starting point. Starting points were counterbalanced across subjects and across the four days of test driving for each subject. The eight parts of the loop differed in the number of lanes and the speed limit, using all combinations of speed limit (35 mi/h [56 km/h] or 55 mi/h [89 km/h]) and number of lanes (two lanes or four lanes) twice. Test Procedure Results While the increase in crash rate after partial sleep deprivation did not reach statistical significance, it should be noted that a relatively small sample of young, healthy subjects was used and that the effects of continuous driving for long periods of time were not assessed.
Although crashes were the most dramatic variable, other measures of driving performance were affected by sleep deprivation. Lateral placement variance (the square of the standard deviation of lateral placement averaged over 100-meter segments) increased with progressive sleep deprivation, as shown in figure 2. Lane excursions (drivers exceeding their lane boundaries when not making a lane change) also increased with progressive sleep deprivation, as shown in figure 3. Speed increased with progressive sleep deprivation in the slower (35 mi/h) zones but not the faster (55 mi/h) zones, as seen in figure 4.
Three variables (lateral placement variance, number of lane excursions, and steering wheel position variance) were significantly correlated with crashes. These variables were simultaneously entered as independent variables into a regression equation with number of crashes as a dependent variable. This yielded a significant multiple R of 0.927, accounting for 86 percent of the variance in crashes. A stepwise regression procedure was then performed, and the first step (when lateral placement was entered into the equation) achieved a significant multiple R = 0.898, accounting for 81 percent of the variance in crashes. The stepwise procedure stopped at this point because entering either of the remaining two driving variables would not have significantly added to the predictive power of the regression equation. Discussion Quantitative extrapolations cannot be made from the simulator data because the simulator allowed subjects to resume driving immediately after an off-road crash or a collision with another vehicle. Thus, the absolute number of recorded accidents far exceeds what could occur in an actual 40-minute drive. Although this aspect of simulation may be viewed as unrealistic, it increases measurement sensitivity and provides the ability to detect low-probability events. The findings of this study lend support to the value of rumble strips or other devices to warn, alert, or awaken drivers as they approach the road edge. When a subject had an off-road crash, a loud crash noise coincided with the event. This noise was sufficient to alert or awaken the subject, after which the subject would continue to drive. However, with total sleep deprivation, even after being awakened after the first off-road crash, several subjects repeatedly had additional off-road crashes. This finding suggests that highway design aids, such as rumble strips, that use noise or vibrations to alert/awaken drivers who are exceeding lane boundaries may be very effective initially but might not be sufficient to maintain the alertness of drivers who continue to drive without rest. Preliminary countermeasures for drowsy driving problems were identified and defined in this effort. Clearly, after a crash occurs, it is too late for interventions to warn drivers of their dangerously drowsy states. Similarly, lane excursions have such significant safety implications that such warnings are also too late to be of use. Lateral placement changes, however, can be calculated as the driver operates the vehicle. Because lateral placement variance was a significant and robust predictor of crashes, this driving performance measure shows significant promise as a method for early detection of sleepiness that may lead to crashes. A warning could be provided to drivers when this measure exceeds a certain threshold to alert them to potential safety problems. Further Research As a result of sleep deprivation studies, methods of predicting critical performance failures due to sleepiness are being developed. Knowledge of the variables leading up to a crash brings us one step closer to the development of highway engineering and in-vehicle drowsy driver warning systems. Such systems can prevent the drowsy driver from endangering themselves and others, thereby improving highway safety. A preliminary neural net analysis using the data collected in this effort is underway. The neural net is being used as a pattern recognizer, and both driving performance data that precipitated crashes and performance data that did not lead to crashes will be fed into the neural net. Theoretically, the neural net will be able to differentiate between normal patterns of driving performance data and to recognize patterns in the driving performance data that had a high likelihood of resulting in a crash. If the patterns can be identified, this demonstration will lend strong support for the development of a neural net in-vehicle-based system for detecting and warning drowsy drivers of potential danger and will hopefully prevent fatigue-related crashes. References Dr. Robert D. Peters is a senior research psychologist at Science Applications International Corporation (SAIC) in McLean, Va. He conducted research in the FHWA Human Factors Laboratory at the Turner-Fairbank Highway Research Center (TFHRC) under a support services contract. He has a doctorate in human factors psychology from George Mason University. Esther Wagner is a research psychologist in the National Highway Traffic Safety Administration's Research and Evaluation Division. She formerly conducted research in the FHWA Human Factors Laboratory at TFHRC under a support services contract. She has a master's degree in human factors psychology from George Mason University. Elizabeth Alicandri is an engineering research psychologist and the manager of the Human Factors Laboratory in the Traffic and Driver Information Systems Division of FHWA's Office of Safety and Traffic Operations Research and Development at TFHRC. She has participated in human factors highway safety research since 1984. She has a bachelor's degree in psychology from Georgetown University and is currently a master's degree candidate in traffic engineering at the University of Maryland. Dr. Jean E. Fox was research assistant at TFHRC during this study. She has a doctorate in human factors psychology from George Mason University. Dr. Maria L. Thomas, Dr. David R. Thorne, Helen C. Sing, and Sharon M. Balwinski are sleep researchers at Walter Reed Army Institute of Research, Division of Neuropsychiatry, Department of Behavioral Biology in Washington, D.C.
Articles & Departments
|
TFHRC Home | FHWA
Home | Feedback
United States Department of Transportation - Federal Highway Administration |