Minnesota Researchers Tackle Driver Fatigue Detection and Warning
IVsource.net
1 April 2002

Joining the growing number of researchers in the complex field of driver fatigue monitoring, investigators at the University of Minnesota have achieved some successes with a system that detects symptoms of fatigue and warns drivers accordingly.

Joining the growing number of researchers in the complex field of driver fatigue monitoring, investigators at the Department of Computer Science and Engineering at the University of Minnesota, with funding from the ITS Institute, are working on a system to detect symptoms of fatigue.  Setting them apart from many of their other colleagues, however, are the further steps they take to produce timely warnings that could prevent a crash.

Currently, most researchers in the field of driver fatigue research focus primarily on the detection phase, leaving effective warning approaches to be defined by other scientists.

A new report by Harini Veeraraghavan and Professor Nikolaos Papanikolopoulos, Detecting Driver Fatigue Through the Use of Advanced Face Monitoring Techniques, presents and tests an approach for real-time fatigue detection.

The system described in the report is based on detecting, through the use of a video camera, indicators of a driver's oncoming fatigue. One such symptom of fatigue is the "micro-sleep"—a short period in which a person loses consciousness for three to four seconds. 

Several other companies and research institutes are also developing driver fatigue detection systems based on video monitoring.  

The Minnesota researchers' approach uses a video camera to capture a continuous stream of images of a driver's face.  From the stream, their system uses skin color information and "blob" analysis to locate the face and eyes of the driver, then monitors how long the driver's eyes are open or closed.  If the eyes remain closed for a specified period of time (three to four seconds), the system determines that the driver is fatigued and issues a warning signal.

The researchers tested the system on 20 subjects with differing skin color, facial hair, and gender.  Overall, the system performed well, detecting the blinks of drivers and giving no false alarms.

[Top]

For More Information ...

... and to download the entire report, go to www.its.umn.edu/news/2001/fatigue.html , or to www.cts.umn.edu/pdf/CTS-01-05.pdf.

[Top]