Driver Fatigue Detection Based on Saccadic Eye Movements

Abstract

The correct determination of driver's level of fatigue has been of vital importance for the safety of driving. There are various methods, such as analyzing facial expression, eyelid activity, and head movements to assess the fatigue level of drivers. This paper describes the design and prototype implementation of a driver fatigue level determination system based on detection of saccadic eye movements. Driver's eye movement speed is used to assess driver's fatigue level. The information about eyes is obtained via infrared led camera device. Movements of pupils were recorded in two driving scenarios with different traffic density. In the first scenario, the traffic density was set to low while the second scenario was based on high density and aggressive traffic. Based on the movements of pupils, the data on saccadic eye movement was analyzed to determine fatigue level of the driver. Acceleration, speed, and size of pupils at both traffic scenarios were compared with data mining techniques, such as segmentation adaptive peak, entropy, and data distribution analyses. Significantly different levels of fatigue were found between the tired and vigorous driver for the different types of scenarios.