自動車運転時の眠気の予測手法についての研究 : 第2報, 眠気変動の予測と居眠り警報タイミングの決定

概要

論文の詳細を見る
The usefulness of various physiological and other indexes for prediction of changes in driver sleepiness during driving of a car was examined in automobile simulator experiments. Based on the results presented in our first report, a multiple regression model for prediction of the sleepiness independent of individual differences was developed. Using the relative energy (α/β) of an electroen-cephalogram at the posterior-midline (Pz), blink duration, and heart rate as explanatory variables, we obtained a mean multiple correlation coefficient of 0.697. We then developed a practical doze prevention system for drivers. Blinking was selected as the physiological parameter since blinking is measurable without direct contact. The practical prediction indexes of sleepiness were the blink duration and the rate of long blinks. The mean correlation coefficient by the prediction model that adopted the rate of long blinking as an explanatory variable was 0.705, while that obtained using the model with blink duration as an explanatory variable was 0.710. Through use of both of prediction models in combination, subjects' (N=10) sleepiness was predicted before reached a level considerd to be dangerous for in driving of a car. Based on these results, a flow chart of the process from the prediction of sleepiness to the sounding of an alarm for drivers was constructed.
社団法人日本機械学会の論文
1997-09-25