A neuro-fuzzy control alarm on momentum of driving behavior for detecting and combating driver fatigue

Abstract

Car accidents are one of the major causes of death in modern society and driver fatigue is identified as one major risk factor. In this study, a neuro-fuzzy detecting and combating fatigue system, a possible more economical solution for detecting and combating driver fatigue, was developed. First, we surveyed and selected adaptive tendency indices (i.e. relative strength index (RSI), stochastic oscillators (KDL), and moving average convergence divergence (MACD)) based actual driving performance to predict driver fatigue and found that RSI, KDL, and MACD of vehicle speed indicated significantly different momentum before and after the occurrence of fatigue. RSI and MACD were also significant in lateral position, but were insignificant in steering wheel angle. Second, a neuro-fuzzy technology was used to integrate multi-tendency indices for exactly predicting fatigue and control alarm for combating fatigue. The experiment results showed that the neuro-fuzzy system could efficiently alert subjects for keeping concentration on traffic conditions. The alarm reducing some symptoms of fatigue is marginally significant.