Driver fatigue: the importance of identifying causal factors of fatigue when considering detection and countermeasure technologies

Transportation Research Part F 12 (2009) 218–224 Driver fatigue: The importance of identifying causal factors of fatiguewhen considering detection and countermeasure technologies Department of Psychology, Old Dominion University, Norfolk, VA 23529, United States Driver fatigue is an ill-defined term in the literature. It has been broadly used to refer to a wide range of driver states, each with different causal mechanisms. Technologies currently exist which enable detection of driver fatigue and interventions that have the potential to dramatically reduce crash probability. The successful implementation of these technolo- gies depends on the cause and type of fatigue experienced. Sleep-related (SR) forms of dri-ver fatigue result from accumulated sleep debt, prolonged wakefulness or troughs in thecircadian rhythms. SR fatigue is resistant to most intervention strategies. Conversely, tech-nologies for detecting and countering task-related (TR) fatigue (caused by mental overload or underload) are proving to be effective tools for improving transportation safety. Meth- ods of detecting and counteracting the various forms of driver fatigue are discussed.
Emphasis is placed on examining the effectiveness of existing and emerging technologiesfor combating TR forms of driver fatigue.
Ó 2008 Elsevier Ltd. All rights reserved.
Fatigue is a multidimensional construct that has been difficult for researchers to define ). In fact, fatigue, sleepiness and drowsiness are often used synonymously (Fatigue and sleep are con-tributing factors in thousands of crashes, injuries and fatalities annually ). Severalresearchers have reviewed technological devices for the detection and countering of driver fatigue (To our knowledge, no attention has been given to differentiating tech-nology by subcategorizing driver fatigue based on the causal factors of the fatigue. This distinction is important when eval-uating the effectiveness of these technological devices. Driver fatigue can be subcategorized into sleep-related (SR) and task-related (TR) fatigue based on the causal factors contributing to the fatigued state. Sleep deprivation, extended duration ofwakefulness and time of day (circadian rhythm effect) affect SR fatigue. Certain characteristics of driving, like task demandand duration, can produce TR fatigue in the absence of any sleep-related cause.
Certain detection technologies may be helpful in determining both types of fatigue, but technology aimed at increasing alertness may only be suitable for countering TR fatigue. Research has shown that among all the self-initiated strategies thatdrivers use (i.e. cranking up the AC, blasting the radio, rolling down the window), only naps and caffeine produce a reductionin driver fatigue, specifically in the context of SR fatigue ). Thisreview will focus on examining the effectiveness of driver fatigue technologies in the context of these different causal factors.
SR fatigue can be caused by circadian rhythms, sleep deprivation and sleep restriction. Sleep/wake patterns follow the body’s natural circadian rhythm or internal clock, which drives humans to sleep during the night and be awake during * Corresponding author. Tel.: +1 757 683 4439; fax: +1 757 683 1087.
1369-8478/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.trf.2008.11.005 J.F. May, C.L. Baldwin / Transportation Research Part F 12 (2009) 218–224 the day. The circadian rhythm also produces an alertness dip in the early afternoon during which people are sleepier Performance decrements are evident during the troughs in the circadian rhythm. For instance, an increased amount ofsleep-related crashes occur between 2 and 6:00 am and also between 2 and 4:00 pm, which correspond to these troughs(Circadian effects have also been demonstrated during a driving simulator task (Speed deviation (standard deviation of driver’s average speed) varied significantly as a result of time of day, with thegreatest variability occurring at 6:00 am, 2:00 pm and 2:00 am. SR fatigue is also influenced by homeostatic factors, such asthe duration of wakefulness and sleep deprivation. Performance becomes worse the longer a person remains awake. Sleeprestriction, or not obtaining adequate sleep will also result in increased sleepiness and a decline in performance.
In an effort to evaluate the progression of performance decline in response to various levels of sleep deprivation, tested participants on the psychomotor vigilance task (PVT). The PVT is a reactiontime task that requires participants press a button as soon as they see numbers appear on a display. Participants com-pleted this task after 0, 2 (from 3 am to 5 am), 5 (from 1:30 am to 6:30 am) or 8 (from midnight to 8 am) hours ofsleep. The PVT was performed at 10 am. All measures of the PVT, including number of lapses, slowest reaction time,fastest reaction time, time on task decrement and median reaction time improved as hours of sleep increased. Resultsfrom a 40-h sleep deprivation study showed that PVT performance decreased (lapses increased in frequency and reactiontime increased) as the homeostatic pressure to sleep (or time awake) increased ().
TR fatigue is caused by the driving task and driving environment. as well as suggest that driver fatigue can be produced by active or passive TR fatigue. Active fatigue is themost common form of TR fatigue that drivers experience ). relate active fa-tigue to mental overload (high demand) driving conditions and passive fatigue with underload conditions. Examples of hightask demand situations include high density traffic, poor visibility, or the need to complete an auxiliary or secondary task (i.e.
searching for an address) in addition to the driving task. Passive fatigue is produced when a driver is mainly monitoring thedriving environment over an extended period of time when most or the entire actual driving task is automated. Passive fa-tigue may occur when the driving task is predictable. Drivers may start to rely on mental schemas of the driving task whichresults in a reduction in effort exerted on the task Underload is likely to occur when the roadway ismonotonous and there is little traffic.
Most studies of driver fatigue focus on sleep deprivation or circadian rhythm effects, but require drivers to perform driv- ing tasks during monotonous, highway conditions. This confounds the effects of SR and TR fatigue. Regardless, it is clear thatdriver fatigue does produce performance decrements in driver simulation and on-road driving tasks. illustrates thethree types of fatigue, their causes, consequences and interactions.
conducted a driving simulator study where participants completed a 20 min drive every 3 h between 8:00 am and 8:00 pm after either an 8 h night of sleep or a night of complete sleep deprivation. The driv-ing simulator task also included a secondary reaction time task. Results showed that lane position variability was greater inthe sleep deprivation condition and increased across each trial. Like their previous study, sleep deprived drivers drove closerto the centerline of their lane than non-sleep deprived drivers. The standard deviation of speed was greater for sleep de-prived drivers, and was worse at 8:00 am and 2:00 pm, showing a circadian effect on performance. Mean reaction timewas greater in the sleep deprived condition and improved throughout the day regardless of condition.
tested the effects of sleep restriction on a real highway. Participants slept either 8.5 h or 2 h in the laboratory, and then drove on a straight highway with light traffic and fair weather in a car outfitted with additional pas-senger controls. During the on-road portion of the experiment, a researcher accompanied the driver, prepared to take overthe driving task if necessary. Driving sessions were 105 min long and occurred approximately 2 h apart with a total of 5 ses-sions. After each driving session, participants completed subjective sleepiness scales and a 10 min reaction time test. Partic- J.F. May, C.L. Baldwin / Transportation Research Part F 12 (2009) 218–224 ipants in the sleep restricted condition exhibited significantly more inappropriate line crossings, increasing the risk of suchan action by 8 times that of rested drivers. Mean reaction time and subjective sleepiness were both greater for the sleep re-stricted participants.
tested the effects of monotonous environments on driving performance during a simulated driving task and revealed poorer performance in the monotonous condition, as indicated by a greater number of large steer-ing wheel movements (over corrections). Increased automation of the driving task (e.g., using cruise control) during pro-longed driving and the low task demands of monotonous roadway conditions result in passive fatigue ).
Although the same types of performance decrements may be seen for both SR and TR fatigue, these causal factors must be distinguished when considering technology designed to detect or counter driver fatigue. Devices for detecting driverfatigue may be acceptable for both types of fatigue, but countermeasures will only be effective for combating fatigue.
In addition, it is critical to distinguish between active and passive fatigue. Adding an additional task for the driver to per-form like a reaction time task, could improve performance during passive fatigue. There are, in fact, commercially avail-able devices that function this way. However, an additional task would be detrimental to a driver experiencing TR activefatigue. The following section addresses driver fatigue technology. Devices are grouped according to those intended forfatigue detection, crash prevention and fatigue reduction. These devices are evaluated for their effectiveness in detectingor countering SR and TR fatigue.
Detection and warning technologies use measures that are sensitive to physiological and performance changes in fatigue, such as eye movements, head-nodding and steering performance. The goal of these devices is to warn the driver of their fa-tigue so that they can stop driving and rest. Additional research must determine the effectiveness of these devices in thecontext of different causal factors, such as SR or TR fatigue. However, as SR fatigue may more likely lead to drivers actuallyfalling asleep, these devices may be best suited for identifying SR fatigue.
The PERCLOS system calculates the amount of eyelid closure over the pupil based on video monitoring of the eyes. It mea- sures this in 1–3 min intervals and derives an index of fatigue. Specifically, the algorithm calculates the proportion of timewithin 1 min blocks that the eyelid covers 80% of the pupil (). This system has been validated in on-roaddriving studies (as well as with the PVT PER-CLOS values correlated with lane departures and subjective sleepiness in sleep deprived drivers. PERCLOS values also corre-lated with PVT lapses such that participants with higher PERCLOS values exhibited more lapses. The PERCLOS system hasbeen tested against EEG algorithms, eye blink software, and head-nodding technology and was found to be superior com-pared to the other techniques in detecting fatigue and was highly correlated with performance decrements in the PVT measured eye closure, reaction time and subjective sleepiness during a 3 h driving simulator task in completely sleep deprived and fully rested participants. Sleep deprived participants had a higher frequency of lapses andlonger reaction times, higher PERCLOS values and greater subjective sleepiness. These measures also increased over thedrive.
This system is currently used within the driver fatigue monitor DD850 (Attention Technologies, ) used for fatigue monitoring in commercial truckers. The driver fatigue monitor gives the driver feedbackwith audible alerts and visually presents information on the duration of the eye closures and distance driven during thattime.
As a driver starts falling asleep, his/her head starts to nod as muscles relax The No-NAP ) fits over the driver’s ear and monitors head position. When the head nods down, the driver receives a buzz inthe ear to wake him or her up. This type of technology would be most efficient in detecting the onset of sleep, as the head-nod phenomenon appears to be a last cue of sleepiness before a micro-sleep or sleep period occurs (Other head-nodding technologies include the Dozer’s Alarm (and the micro-nod detection system). The Dozer’s alarm was tested during a simulated driving task at night. Results showed that the alarmdid not improve lane tracking performance and did not allow drivers to maintain alertness, but it did detect drowsiness (Although the head-nod technology appears efficient for detecting the onset of sleep, drivers are likelyunfit to drive safely before this point of sleepiness. As such, the head-nod technology warns drivers too late as to their sleep-iness. In addition, there may be a high potential for false alarms as the driver looks around in his/her environment. Falsealarms cause user mistrust and annoyance in a system. This can lead a user not to use the technology or ignore true alarmswhich may result in a crash.
J.F. May, C.L. Baldwin / Transportation Research Part F 12 (2009) 218–224 Deadman switches are designed for the user to continuously press a switch. When the switch is released, it is assumed that the user has become impaired and alarms sound. The release of the deadman switch has been related to longer reactiontimes, and presence of stages 1 and 2 of sleep (). Unfortunately, like the head-nod tech-nology, this type of system, if developed for vehicle use, would be a late-alert system which identifies sleepy drivers onlywhen sleep has set in.
Crash prevention technologies were designed to prevent crashes regardless of the causal factors. However, because crashes which result from running off the road are characteristic of fatigue-related crashes, rumble strips and lane departurewarnings can reduce the crash risk for drivers experiencing both TR and SR fatigue. Collision avoidance warning systems canhelp fatigued drivers (SR and TR) by compensating for their deteriorated reaction times.
Rumble strips are a roadway design option for monitoring weaving and alerting drivers when they drive off the road.
Rumble strips have a grooved design within the pavement that produces a loud noise and vibrations within the car whena driver crosses or drives along the strip. The advantage of rumble strips is that they are available for all drivers. The twotypes of rumble strips constructed today are milled and rolled rumble strips.
Milled rumble strips are constructed on existing asphalt shoulders or new ones, and are concave circular depressions approximately 180 mm wide, 400 mm long and 13 mm deep (). Milled rumble strips are preferred because theyproduce more noise and vibrations than the other types ). Rolled rumble strips can only be installed in new orreconstructed shoulders while the asphalt is hot. They are constructed by creating rolls of excess asphalt above the flat sur-face of the shoulder and are narrower than the milled rumble strips and result in a shallower tire drop thus producing lessnoise and vibration than milled rumble strips Center line rumble strips have resulted in a reduction ofcrashes along rural two-lane roads ). Rumble strips have significantly reduced the amountof run off the road crashes in New York ().
Lane drifting or departure technology receives information from a video camera that monitors the road ahead and graphs the boundaries of the lane. An alarm sounds when a driver veers out of the lane. SafeTRAC issuch a device. SafeTRAC warns drivers if they begin to drift out of their lane without using their turn signal. This device alsocalculates a continuous ‘score’ of performance, and if the score drops, SafeTRAC generates an alert recommending rest(). In this way, SafeTRAC not only detects lane departures, but lane position variability or weav-ing such that performance can be evaluated.
AutoVue is another lane departure warning system manufactured by Iteris () which incorporates the car’s speed, turn signal, and road markings in order to identify unintentional lane departures. Ifthe car drifts out of the lane, a virtual rumble strip sound is emitted to warn the driver. This system is currently installedin the Infiniti M45 and Infiniti FX passenger vehicles. Both the SafeTRAC and the AutoVUE advertise that they have low falsealarm rates, but neither company provide research to support this claim.
2.2.3. Collision avoidance system (CAS) warnings CAS warnings are currently integrated in the production of some vehicles and are intended to alert drivers of potential rear-end or from the side crashes. This type of technology is useful in vigilance conditions where the ability to detect thesecritical events is reduced. CAS warnings can measure time to collision or temporal headway Temporal headway is defined as the time needed for a car to reach the position of the car ahead of it (Other sensors can detect when another vehicle or obstacle comes within a certain distance from the sides ofthe car. demonstrated with a driving simulator task that in both distracted andnon-distracted drivers, early collision avoidance warnings have been shown to reduce collisions by 80% as compared to driv-ers who did not receive a warning. CAS warnings in this study also reduced collision velocity and minimum time to collision,demonstrating a reduction in crash severity. The collision avoidance warning consisted on both an auditory tone and a visualicon displayed above the instrument panel. found that crashes were reduced in driv-ers exhibiting task-related fatigue when presented with an auditory warning prior to a potential head-on crash in a drivingsimulator task.
Technology designed to reduce fatigue or improve performance is suitable for drivers experiencing TR fatigue. Technolog- ical countermeasures include the use of automation in vehicles to reduce task load as well as interactive technology designedto increase engagement of the driver to prevent passive TR fatigue.
J.F. May, C.L. Baldwin / Transportation Research Part F 12 (2009) 218–224 Automation refers to transferring control of a system or task (in this case, a component of driving such as speed control) from the operator to the vehicle. A recent study examining TR fatigue indicated that lane keeping performance was improvedwhen drivers had cruise control activated Adaptive automation can linkdriver fatigue and task load in an effort to optimize driving performance (). The purpose of adaptiveautomation is to modify the task load of the driver depending on the driver’s state.
described two adaptive automation systems. The generic intelligent driver support system (see ) utilizes a computerized scheduling system which analyzes the mental and visual workload of the driver,prioritizes information messages to send to the driver and decides how to present the information. If the driver is under highvisual workload, the system may present messages verbally or by tactile sensation. Although this system does not specifi-cally address fatigue, it does attempt to reduce driver workload. By reducing driver workload, active TR fatigue may bealleviated.
The SAVE system ) integrates a monitoring unit, an automatic control device and a warning sys- tem. The monitoring unit detects fatigue, intoxication and unresponsive drivers. The automatic control device stops the caron the side of the road if the driver is not responding to warnings. The warning system informs the driver of his or her unsafecondition and also informs surrounding traffic and a traffic control center.
Active or adaptive cruise control is an advancement of cruise control which automatically adjusts the car’s speed in re- sponse to the speed of the vehicle ahead of the driver (). This typeof adaptive automation helps drivers in heavier traffic, but may not be beneficial on straight roadways with little traffic).
The S.A.M.G-3, steering attention monitor () monitors normal corrective steering movements, defined by the micro-corrections that drivers must constantly make to maintain performance. An alarmsounds when normal steering movements cease. In a form of adaptive automation, there is an option to allow the system toturn off cruise control when steering wheel movements cease. Other steering monitors, such as the ZzzzAlert Driver FatigueWarning System and the TravAlert early warning system also monitor corrective steering movements and produce audiblealerts if the driver stops steering, but they do not attempt to modify the driver’s task load or take over any driving systems).
believed that giving drivers challenging secondary tasks to perform when they became tired would increase their alertness. They used a game box called ‘‘Car Mate” that was an auditory and verbal interface in thecar designed to allow the driver to play 12 different games. Results showed that driving with the game box produced fewerincidents and accidents, as well as a lesser decline in vehicle control. Participants reported less drowsiness and fewer epi-sodes of sleep in the game box condition.
The Interactive Knight–Warrior Sleep Alarm is activated by the driver, presumably when the driver feels fatigued. The system will periodically emit a minor auditory alarm at a specified time interval. The driver has 1–3 s to turn offthe alarm; otherwise a secondary wake-up siren will sound in an effort to wake the driver up.
Of all of these technologies, the crash prevention devices may be the most feasible to implement in terms of incorporating them into vehicle construction or roadway design. Lane departure warnings and rumble strips are currently in use to reduceoff the road crashes. Collision avoidance warning systems are also installed in some vehicles to assist in reducing rear-endand side crashes. All three types of technology are potentially beneficial for both SR and TR fatigue.
Classifying driver fatigue into SR and TR (active and passive) categories based on the causal factors involved will lead to improved development and implementation of fatigue countermeasures. TR active fatigue stemming from high task loaddriving condition requiring sustained attention and prolonged driving will benefit from increased automation and in-vehicletechnologies that offset driver workload. Conversely, further automation of the driving task will exacerbate TR passive fati-gue caused by monotonous conditions and highly familiar roadways. When a driver is susceptible to TR passive fatigue,increasing the novelty and demand of the driving task through interactive technologies such as the CAR MATE can be of ben-efit. SR fatigue, on the other hand, comes about from dips in the circadian rhythm (time of day effects), sleep deprivation andprolonged wakefulness and may therefore be rather impervious to either form of countermeasure. However, either form ofTR fatigue can exacerbate the performance decrements associated with SR fatigue.
SR and both forms of TR fatigue may be measured similarly. Each form of fatigue has been shown to cause decrements in driving performance and reductions in visual scanning patterns and eye movements. However, performance patterns asso-ciated with each form are frequently confounded in the existing literature by examining them in combination rather thanisolation. Future research should further examine the potential performance differences between SR and TR fatigue formsin isolation as well as in combination. For example, SR fatigue may result in a different signature of impairment in isolationversus in combination with either active or passive TR fatigue. If such a fatigue profile can be reliably detected, then adaptiveforms of fatigue countermeasures might be implanted that target the specific interaction with the specific type of fatigue J.F. May, C.L. Baldwin / Transportation Research Part F 12 (2009) 218–224 Table 1The effectiveness of driver fatigue technologies in detecting or combating TR and SR fatigue.
Crash preventionLane departure warning systems being experienced. presents a summary of the countermeasures proposed to be beneficial for each type of fatigue.
Several studies have shown that a nap and/or caffeine are the only effective countermeasures to SR fatigue ().
Crash prevention technologies, such as lane departure warnings and collision avoidance warning systems, are likely to be helpful in both forms of TR and SR fatigue. In both types of fatigue, lane position variability and the potential for lane depar-tures increase. Off the road crashes are the most common type of crashes from falling asleep at the wheel ).
Rumble strips are helpful in alerting drivers to off the road incidents, and are being installed more frequently on majorhighways.
In conclusion, future research testing measures of fatigue and fatigue-related technology must distinguish between dif- ferent forms of fatigue. Countermeasure technologies for driver fatigue will be more or less effective depending on the typeof fatigue the driver is experiencing. These technologies may be useful especially in the truck driving industry to help detectSR fatigue to indicate a need to take a break, or to help manipulate task load to better engage the driver. Future technologiesthat integrate several measures of fatigue (such as PERCLOS and lane departure warnings) might be more reliable and effec-tive at detecting driver fatigue and reducing crashes associated with fatigue.
AssistWare Technology. (2005). Third generation safetrac released. Retrieved August 22, 2005, from Barton, T. (2003). On the safe side. Overdrive.
Ben-Yaacov, A., Maltz, M., & Shinar, D. (2002). Effects of an in-vehicle collision avoidance warning system on short- and long-term driving performance.
BMW World. (n.d.). BMW technology: Active cruise control. Retrieved August 25, 2006, from Boivin, D. B. (2000). Best practices compendium of fatigue countermeasures in transport operations (No. TP 13620E): Transport Canada.
Brown, I. D. (1994). Driver fatigue. Human Factors, 36(2), 298–314.
CARRS-Q. (2002). Inventors forum. Invention: Interactive knight-warrior sleep alarm. Retrieved August 22, 2005, from Desmond, P. A., & Hancock, P. A. (2001). Active and passive fatigue states. In P. A. Hancock & P. A. Desmond (Eds.), Stress workload and fatigue (pp. 455–465).
Mahwah: Lawrence Erlbaum Associates, Inc.
Dinges, D. F., & Grace, R. (1998). PERCLOS: A valid psychophysiological measure of alertness as assessed by psychomotor vigilance (No. FHWA-MCRT-98-006). US Department of Transportation: Federal Highway Administration.
Dinges, D. F., Mallis, M., Maislin, G., & Powell, J. W. (1998). Final report: Evaluation of techniques for ocular measurement as an index of fatigue and as the basis for alertness management (No. DOT HS 808 762). National Highway Traffic Safety Administration.
Electronic Safety Products. (n.d.). S.A.M.G-3-steering attention monitor. Retrieved 2005, September 1, from .
Funke, G. J., Matthews, G., Warm, J. S., Emo, A., & Fellner, A. N. (2005). The influence of driver stress, partial-vehicle automation, and subjective state on driver performance. Paper presented at the human factors and ergonomics society 49th annual meeting, Orlando, FL.
George, C. F. P. (2005). Driving and automobile crashes in patients with obstructive sleep apnoea/hypopnoea syndrome. Sleep Diagnosis and Therapy, 1(1), Gimeno, P. T., Cerezuela, G. P., & Montanes, M. C. (2006). On the concept and measurement of driver drowsiness, fatigue and inattention: Implications for countermeasures. International Journal of Vehicle Design, 42(1/2), 67–86.
Global Infotech Pune. (n.d.). Retrieved June 14, 2006, from .
Graw, P., Krauchi, K., Knoblauch, V., Wirz-Justice, A., & Cajochen, C. (2004). Circadian and wake-dependent modulation of fastest and slowest reaction times during the psychomotor vigilance task. Physiology and Behavior, 80, 695–701.
Hancock, P. A., & Verwey, W. B. (1997). Fatigue, workload and adaptive driver systems. Accident Analysis and Prevention, 29(4), 495–506.
Hartley, L., Horberry, T., Mabbott, N., & Krueger, G. P. (2000). Review of fatigue detection and prediction technologies. National Road Transport Commission.
Haworth, N. L., & Vulcan, P. (1991). Testing of commercially available fatigue monitors (No. 15). VIC Roads and The Victorian Road Freight Transport Industry Horne, J., & Reyner, L. (2001). Sleep-related vehicle accidents: Some guides for road safety policies. Transportation Research Part F, 4, 63–74.
Jewett, M. E., Dijk, D., Kronauer, R. E., & Dinges, D. F. (1999). Dose–response relationship between sleep duration and human psychomotor vigilance and subjective alertness. Sleep, 22(2), 171–179.
Johns, M. W. (1998). Rethinking the assessment of sleepiness. Sleep Medicine Reviews, 2(1), 3–15.
Knipling, R. R., & Wang, J. S. (1994). Crashes and fatalities related to driver drowsiness/fatigue, NHTSA Research Note. Washington, DC: National Highway J.F. May, C.L. Baldwin / Transportation Research Part F 12 (2009) 218–224 Kozak, K., Curry, R., Greenberg, J., Artz, B., Blommer, M., & Cathey, L. (2005). Leading indicators of drowsiness in simulated driving. Paper presented at the human factors and ergonomics society 49th annual meeting, Orlando, FL.
Lee, J. D., McGeehee, D. V., Brown, T. L., & Reyes, M. L. (2002). Collision warning timing, driver distraction, and driver response to imminent rear-end collisions in a high-fidelity driving simulator. Human Factors, 44(2), 314–334.
Lenne, M. G., Triggs, T. J., & Redman, J. R. (1997). Time of day variations in driving performance. Accident Analysis and Prevention, 29(4), 431–437.
Lenne, M. G., Triggs, T. J., & Redman, J. R. (1998). Interactive effects of sleep deprivation, time of day and driving experience on a driving task. Sleep, 21(1), Mallis, M. M. (1999). Evaluation of techniques for drowsiness detection: Experiment on performance-based validation of fatigue-tracking technologies.
Unpublished Doctoral Thesis, Drexel University, Philadelphia, PA.
Marsden, G., McDonald, M., & Brackstone, M. (2001). Towards an understanding of adaptive cruise control. Transportation Research Part C, 9, 33–51.
May, J. F., Baldwin, C. L., & Parasuraman, R. (2006). The effectiveness of auditory collision avoidance warnings in reducing fatigue-related crashes. Paper presented at the proceedings of the human factors and ergonomics society 50th annual meeting, San Francisco, CA.
Maycock, G. (1996). Sleepiness and driving: The experience of UK car drivers. Journal of Sleep Research, 5, 229–237.
Michon, J. A. (1993). The generic intelligent driver support: A comprehensive report on GIDS. CRC Press.
Monk, T. H. (1991). Circadian aspects of subjective sleepiness: A behavioral messenger? In T. H. Monk (Ed.), Sleep, sleepiness and performance (pp. 39–63).
Chichester; New York: John Wiley and Sons.
NHTSA. (2006). Traffic safety facts 2004: A compilation of motor vehicle crash data from the fatality analysis reporting system and the general estimates Ogilvie, R. D., Wilkinson, R. T., & Allison, S. (1989). The detection of sleep onset: Behavioral, physiological, and subjective convergence. Sleep, 12(5), 458–474.
Pack, A. I., Pack, A. M., Rodgman, E., Cucchiara, A., Dinges, D. F., & Schwab, C. W. (1995). Characteristics of crashes attributed to the driver having fallen asleep. Accident Analysis and Prevention, 27(6), 769–775.
Perrillo, K. (1998). The effectiveness and use of continuous shoulder rumble strips. Albany, NY: Federal Highway Administration.
Persaud, B. N., Retting, R. A., & Lyon, C. A. (2004). Crash reduction following installation of centerline rumble strips on rural two-lane roads. Accident Analysis Philip, P., Sagaspe, P., Moore, N., Taillard, J., Charles, A., Guilleminault, C., et al (2005). Fatigue, sleep restriction and driving performance. Accident Analysis Philip, P., Taillard, J., Moore, N., Delord, S., Valtat, C., Sagaspe, P., et al (2006). The effects of coffee and napping on nighttime highway driving. Annals of Reyner, L. A., & Horne, J. A. (1998). Evaluation of ‘in-car’ countermeasures to sleepiness: Cold air and radio. Sleep, 21(1), 46–50.
Reyner, L. A., & Horne, J. A. (2000). Early morning driver sleepiness: Effectiveness of 200 mg caffeine. Psychophysiology, 37, 251–256.
Reyner, L. A., & Horne, J. A. (2002). Efficacy of a ‘functional energy drink’ in counteracting driver sleepiness. Physiology and Behavior, 75, 331–335.
Thiffault, P., & Bergeron, J. (2003). Monotony of road environment and driver fatigue: A simulator study. Accident Analysis and Prevention, 35(3), 381–391.
Verwey, W. B., & Zaidel, D. M. (1999). Preventing drowsiness accidents by an alertness maintenance device. Accident Analysis and Prevention, 31, 199–211.
Wierwille, W. W., Lewin, M. G., & Fairbanks, R. J. (1996). Final reports: Research on vehicle-based driver status/performance monitoring, Part III (No. DOT HS 808 640). National Highway Traffic Safety Administration.
Young, M. S., & Stanton, N. A. (2004). Taking the load off: Investigations of how adaptive cruise control affects mental workload. Ergonomics, 47(9),

Source: http://trafiksikkerhedsforskning.dk/images/Driver%20fatigue_artikel_4.pdf