A description is given of a portable eye-marker camera and a specially designed stabilization unit that may be used to record drivers' eye movements. A pilot study was conducted that showed that drivers' eye movements are closer in toward the vehicle under night driving than day driving. The pilot study also pointed to the need to examine the calibration accuracy of the system. In phase 1 of the calibration experiment, seven drivers served in a 2 x 2 x 2 within- subjects design. The three independent variables were head movements (calibration before vs calibration after), distance (30 vs 60ft.) and sessions (day 1 vs day 2). Sessions were investigated to see if subjects can be calibrated with the same accuracy on different days. In phase 11 of the experiment, calibration accuracy was measured before and after the subjects drove an automobile. Calibration accuracy was measured by having the drivers trace a matrix of targets that covered the field of view of the camera. Error was defined as the distance from the center of the target to the center of the eyespot. Analyses of variance showed that head movements, distance, and sessions had no significant effect on calibration accuracy. The effect of driving on calibration accuracy was small. The average calibration error for all subjects under all conditions was 1 degree. Potential uses of the eye-marker camera in driving research are discussed. /author/.
Abstract