Emotional state detection on mobility vehicle using camera: Feasibility and evaluation study

Abstract

This paper demonstrates the feasibility of using a camera to estimate the emotional state embedded on personal mobility two-wheel balanced vehicle. The Heart Rate Variation (HRV) metrics are used to recognise the two states: stressed and relaxed. We proceed into three steps: First, we validate the acquisition chain with camera, comparing the results with a contact sensor in static conditions for the first experiment; second, we obtain the results of the emotional state classification using neural network and support vector machine methods. Then, we improve these results using Boost methods for the classification. Finally, we design a second experiment in quasi-dynamic conditions. Results indicate that using a low-cost camera to estimate the emotional state of an individual is feasible.