Location

Aspen, Colorado, USA

Session

Poster Session 2

Abstract

The research project compared and analyzed physiological and performance data for 13 subjects driving a vehicle simulator. Each subject drove the simulator for morning, afternoon, and late night sessions. These sessions were intended to represent alertness conditions during an “awake” baseline period and the secondary and primary circadian sleep cycle periods. The sessions were approximately one hour, two hours, and two or three hours in length, respectively. With one exception, the subjects had experienced normal sleep the night before the test. Five men and eight women participated, ranging in age from 25 to 59. Physiological data included: real-time PERCLOS (percentage of slow-eye closure over one minute) using an infrared-reflective camera; head position coordinates using an overhead capacitive sensor array; and video of the right front of the subject’s face. Performance data included: vehicle speed, lane departures, lane deviation, and steering/turn signal data. The research manager maintained logs of unusual circumstances such as departing the roadway, falling asleep at the wheel, excessive speeding, etc. Head position data was analyzed and compared to the videos. A multi-element algorithm was developed which captured patterns of head motion found to be characteristic of drowsiness. The algorithm output was compared to roadway departures noted in the research manager’s logs of unusual events. The comparison showed a capability of advance detection of about 87% of driver roadway departures with a false positive rate of about 15%.

Rights

Copyright © 2001 the author(s)

Share

COinS
 
Aug 16th, 12:00 AM

Driver Alertness Detection Research Using Capacitive Sensor Array

Aspen, Colorado, USA

The research project compared and analyzed physiological and performance data for 13 subjects driving a vehicle simulator. Each subject drove the simulator for morning, afternoon, and late night sessions. These sessions were intended to represent alertness conditions during an “awake” baseline period and the secondary and primary circadian sleep cycle periods. The sessions were approximately one hour, two hours, and two or three hours in length, respectively. With one exception, the subjects had experienced normal sleep the night before the test. Five men and eight women participated, ranging in age from 25 to 59. Physiological data included: real-time PERCLOS (percentage of slow-eye closure over one minute) using an infrared-reflective camera; head position coordinates using an overhead capacitive sensor array; and video of the right front of the subject’s face. Performance data included: vehicle speed, lane departures, lane deviation, and steering/turn signal data. The research manager maintained logs of unusual circumstances such as departing the roadway, falling asleep at the wheel, excessive speeding, etc. Head position data was analyzed and compared to the videos. A multi-element algorithm was developed which captured patterns of head motion found to be characteristic of drowsiness. The algorithm output was compared to roadway departures noted in the research manager’s logs of unusual events. The comparison showed a capability of advance detection of about 87% of driver roadway departures with a false positive rate of about 15%.