EEG-Based Emotion Recognition Using Deep Learning Model for Workers Safety

Abstract

Recently, industrial robots and collaborative robots are widely used in industrial sites with the introduction of smart factory. In a human-robot collaboration environment, it is important to ensure the safety of workers above all. This study suggested an EEG-based deep learning model-based worker safety management system that guards employees by identifying their feelings when they perceive risk. We evaluated and examined the performance of the suggested CNN, DNN, LSTM, and CNN-LSTM models in order to determine which deep learning model would work best for EEG-based emotion identification. With 71.3% accuracy while utilizing the SEED dataset as input information, the CNN-LSTM model demonstrated good performance; with 74.4% accuracy, the CNN model demonstrated good performance when using the real gathered data set. The proposed deep learning model has a small number of parameters, a small size, and fast processing time, which is advantages for real-time application.