Conceptual Approach in Determining Fatigueness and Drowsiness Detection Using EEG-Based and Artificial Neural Network

Abstract

Driver fatigue is a major exogenous cause of fatal road accidents and has implications for Malaysian road safety. The major aspect that causes human errors is fatigue/drowsiness due to task-induced factors (environmental circumstance) or attitude/behaviour of the driver (lack of sleep, consumption of alcohol, long driving hours and etc.). Therefore, it is necessary to identify significant index to detect driver fatigue and associate that index with the level of alertness (active or asleep), for road safety and for use by regulatory bodies such as Jabatan Pengangkutan Jalan (JPJ) and Police for advice on driving conditions. This can be carried out by observing the physiological behaviour through the Event-related potentials (ERPs) and electroencephalography (EEG) measures. ERP’s are very small voltage potentials that examine the information processing and characterize the brain structures in response to specific events or stimuli. Studies have shown EEG changes that are time-locked to sensory, cognitive or motor events are the most promising psychophysiological measures for assessing mental process and better indicators of fatigue. Hence, in this research work, to detect the driver fatigue and associate with alertness, it is proposed to develop an adaptive fatigue identification model based on the EEG frequency spindles (alpha and theta waves that reflects the cognitive and memory process). The proposed adaptive model identifies the driver's fatigue by acquiring EEG signals using a suitable EEG recording protocol that distinguishes the brain's perception in response to various environmental changes and driver behavioural aspects. The recorded signals will be analysed to extract discriminant features through cross-correlation techniques and Neurofuzzy algorithm for the classification of level of fatigue.