Canada, U.S. to study fatigue-monitoring technologies

NASHVILLE, Tenn. (Sept. 3, 1999) — As Canada and the United States continue to hone proposed new hours-of-service regulations, driver fatigue remains a hot research topic.

The University of Pennsylvania School of Medicine has just wrapped up a simulator study of six different technologies that claim to detect driver drowsiness. Dr. David Dinges told a U.S. National Transportation Research Board hearing this week that only one of those systems could reliably predict drowsiness-related performance failures.

The system is called Perclose, and it measures the percentage of eyelid closures through a video monitoring system. Dinges called it “astonishingly good,” better than the drivers’ own evaluation of their fatigue level. The other drowsiness detectors only worked on some people some of the time. They included brain wave systems, head positioning system and another eyeblink device.

The Perclose system is one of at least four fatigue-related technologies that will be tested in the real world in a soon-to-be-launched pilot program headed up by the American Trucking Associations Foundation. The program will also test the wrist actograph, onboard data recorders, and the Howard Power Center Steering System, which is supposed to reduce physical fatigue by helping the truck track straight. Bill Rogers with ATA told the NTSB hearing that they may add lane tracking cameras to the list.

The study will be done in co-operation with Transport Canada. The program will study 24 drivers, four at a time. Different hours-of-service scenarios will also be tested in conjunction with the technology. Several trucking companies have volunteered to participate in the study, which is partially funded by the federal government.

In addition, Dinges noted that a University of Pennsylvania study on sleep apnea in truck drivers is just wrapping up. It looks feasible, he said, that there will be a cost-effective way to test drivers for apnea and to treat it much more cheaply than current methods.