Testing of Commercially Available Fatigue Monitors
Monash University Accident Research Centre - Report #15 - 1991
Authors: N. Haworth, P. Vulcan
Full report in .pdf format [2.2MB]
This report presents an evaluation of three commercially available fatigue monitors (the Onguard eye closure monitor, a head nodding monitor - Dozer's Alarm, and an analogue of the Roadguard reaction time monitor) based on laboratory testing.
The devices showed an ability to detect fatigue in some cases but were not able to maintain alertness and thus prevent performance deterioration. There were few instances of a startle response to the presentation of an auditory warning signal by Dozer's Alarm or Roadguard. Equipment faults with the modified version of Onguard made it impossible to assess the frequency of startle responses to the alarm.
This is the third and final report of a programme of research into driver fatigue in truck accidents. This report presents an evaluation of three commercially available fatigue monitors based on laboratory testing. The research programme was commissioned by the Victorian Road Freight Transport Industry Council and funded by the then Road Construction Authority in May, 1988.
The first report, Fatigue in truck accidents (Haworth, Heffernan and Horne, 1989), presented two estimates of the involvement of fatigue in Victorian truck accidents and a review of in-vehicle fatigue countermeasures. The second report was entitled Information for development of an educational program to reduce fatigue-related truck accidents (Haworth and Heffeman, 1989).
Based on the recommendations of the first report, the Onguard eye closure monitor, a head nodding monitor (Dozer's Alarm) and a laboratory simulation of the Roadguard reaction time monitor were tested.
The criteria applied in this evaluation were reliability, threshold and degree of intrusiveness. An examination of the driver fatigue literature showed that eye closure and subsidiary reaction time (the measures used by the eye closure monitor and the Roadguard reaction time monitor) have been found to be valid measures of driver fatigue.
Laboratory testing took place at night. Subjects watched a videotape which showed the driver's eye view of daytime driving around a circular test track. All subjects performed a tracking task which was superimposed on the road scene.
The performance of subjects using the monitors was compared with that of subjects in the control condition. The three questions addressed were
- Was the monitor able to detect the onset of fatigue?
- Did use of the monitor allow the subject to drive for longer without falling asleep?
- Did use of the monitor result in less deterioration in performance?
The answer to the first question was generally "yes" for the Roadguard device and "sometimes" for the other two devices. The answers to the second and third questions were "no".
Subjects who used the Onguard, Head nodding or Roadguard devices did not drive for longer without failing asleep. The results may have differed with a larger group of subjects but this is unlikely based on the lack of clear trends in the four groups of six subjects used in this study.
The reaction times before and after simulated driving and the measures recorded during simulated driving (number of long eye closures, total time eyes closed, tracking error) showed deterioration in performance with time on task but the amount of deterioration was not affected by which monitor was used (or whether a monitor was used).
Ale fatigue monitors were also evaluated by examining critical incidents. Performance just before and just after warning signal was analysed to clarify the immediate effect of the warning. The analyses looked for evidence of
1. An improvement in alertness and vehicle control behaviour following the warning signal (evident for at least five minutes).
2. A startle response to the warning signal (a deterioration in performance of less than one minute in duration).
There was mixed evidence of an improvement in alertness and vehicle control
behaviour following the warning signal. Tracking errors appeared to be somewhat reduced in the two minutes following presentation of the Head nodding alarm. There appeared to be fewer long eye closures following the presentation of the Roadguard alarm. No corresponding reduction in tracking errors was noted, however.
There were few instances of a startle response to the presentation of an auditory warning signal by Dozer's Alarm or Roadguard. Equipment faults with Onguard, which had been modified to enable continuous recording, made it impossible to assess the frequency of startle responses to the alarm.
Subjects rated the head nodding monitor (Dozer's Alarm) as the least annoying of the three monitors and considered it to be the most effective and said they would be more willing to use and purchase this device. However, these trends were not significant when subjected to analysis of variance, possibly because of the small number of subjects in each group. There were no overall correlations between the number of warning signals presented and responses to questionnaire items.
The reliability of the Onguard eye monitor and, to a lesser extent, the Dozer's Alarm is reduced by the tendency to slip from the correct position. Redesign of the devices or more secure attachment could correct this problem, however.
There may be legal issues involved if a driver fell asleep and crashed while using a fatigue monitor. This issue, while outside of our brief, needs to be investigated.
In conclusion, the devices showed the ability to detect fatigue in some cases but were not able to maintain alertness and thus prevent performance deterioration. The need to develop more reliable detectors of driver fatigue still remains. Fatigue detectors can only be used as a wanting to stop driving and rest, not as an aid to prolonging driving.
Sponsors: VicRoads and the Victorian Road Freight Transport Industry Council