Wearable emotion and fatigue evaluation system for automatic control of walking assist device

Abstract

With the growing population of elderlies, the number of senior citizens coping with walking impairments is increasing. It is necessary for elderlies to do proper exercise to prevent systematic disease. Various type of walking assistant devices is invented for rehabilitation or exercise support. However, doing exercise needs high level of motivation to gain effective result. Maintaining a high level of motivation is a critical issue during exercises since it might involve pain, discomfort, or depression especially for elderlies. On the other hand, senior citizens have higher possibility get injury caused by over-exercising. Therefore, the purpose of this research is keeping positive emotion and body condition for senior citizens when they doing exercise using walking assistant device. In this research, we proposed a 3D human condition model to evaluate human emotion and fatigue condition. For emotion recognition, we proposed a Deep Neural Network (DNN) trained by electroencephalogram (EEG) signal and heartbeat signal. For fatigue detection, we proposed a new method by using portable near-infrared spectroscopy (NIRS). In addition, we also found the relationship in living habit and fatigue time. Based on the emotion evaluation system, we can propose a control strategy. By using the control strategy, user can keep doing exercise in high level of motivation while avoiding fatigue damage.