Preliminary Study of Brain-Inspired Model for Multimodal Human Behavior Detection in Social Context

In a near future, robots will permeate our daily life; indeed, they have the potential to proactively support the senior citizen in tedious and different daily tasks (i.e. cleaning, gaming, walking activity, promote socialization). However, to efficiently cooperate with human-beings, robots should have enhanced human–robot interaction capabilities. This work addresses the challenge of designing a robotic model by simulating the modality human beings interact with each other. The first objective of this work concerns the identification of the social cues which correctly describe the user’s emotional and engagement state during the interaction. Based on selected descriptors, a perceptual system has been proposed to detect elderly’s behavior in a social context. The proposed architecture is inspired to the human brain structure as concern the functionalities modules and analogies in the modules’ location. The proposed perceptual system aims at transforming raw data coming from three kinds of sensors (camera, microphone, and laser) into behavioral patterns by mimicking the abstraction evolution which characterizes the consciousness process of the human brain.