Human Computer Interaction using CNN and Fuzzy Inference System (FIS)

Abstract

In the evolutionary world of technology, humanoid robot has gained increased attention in different areas such as security systems, medical systems (psychiatric clinic), entertainment, military applications, home appliances, education etc. Companion robot is one of the major challenges in humanoid robots. Companion robot can play role of companion for children or senior citizens rather than machine assistance. Such robots are expected to exhibit capabilities primarily includes propensity to recognize human emotions with a skill to express its own expressions as sensible as human. This paper aims to present system to recognize human facial expression and respond by generating an expressive robot face. Convolutional Neural Network (CNN) approach is used to recognize human facial expressions. This is accomplished by detecting the occurrence of facial Action Units (AUs) as a subpart of Facial Action Coding System (FACS) which represents human expressions. Real-World Affective Face Database (RAF-DB) has been employed for training the algorithm. Fuzzy inference system is adopted to decide responsive expressions on robotic face simulator.