E-Bot: A Facial Recognition Based Human-Robot Emotion Detection System

Abstract

There are many Emotion Detection Systems which can understand emotion but are only present for monitoring purposes. Emotional Robots currently available predict emotion with low accuracy. Thus there is a requirement for Emotion Detection Robot which can predict emotion of user with high accuracy. The robot should converse with the user based on their emotion. It should be a companion for the user. In this paper we present the development of the E-Bot system, which enables human-robot interaction based on emotion detected from facial recognition. A mobile app is developed to control the robot and guide the chat. The Google Cloud Vision API and Pre-trained Facial Expression Algorithm are explored and the prediction accuracy, sensitivity and specificity of these two approaches are summarised in this paper. Results show that the proposed E-Bot system could be applied to provide affective care for people living alone at home.

Artificial Vision Algorithms for Socially Assistive Robot Applications: A Review of the Literature

“…For instance, refs [ 30 , 36 ] were tested offline. Regarding the computer resources used to recognize facial expressions, two implementations are reported using basic hardware computers [ 34 , 39 ] and computers with graphic cards [ 30 , 31 ]. It is worth noting the use of robotic platforms such as the NAO [ 27 , 28 , 33 , 38 ], R-50 Alice [ 35 , 40 ], Pepper [ 32 ], kiwi [ 44 ], and N-Maria [ 42 ].…”

“…On the other hand, Sridhar et al presented in [ 34 ] the E-bot vision system, which was developed using a Raspberry Pi. The system was used to identify the mood of a person by employing Google Cloud Vision API and facial expression detection (FED); both algorithms used a CNN and internally detected the face through facial feature extraction and the classification of the different moods of a person.…”