A Human Centric Approach to Robotic Control

Abstract

In this paper we present a novel idea for the creation of an intelligent interface that allows the remote control of arbitrarily complex robotics morphologies by translating intuitive human behaviours into purposeful robotic actions. By taking inspiration from human robot interaction, ergonomic principles, and autonomous robotics this paper proposes a human-centric framework for robot control inspired by the current advancements in recurrent neural networks and self-organisation. In particular, we present an integrated approach based on neural networks for input acquisition from human operator and self organisation for the acquisition of robot behaviours. We realise the interface as a kind of intelligent agent connecting the two end points of the system: Human and robot, providing an adaptive and intelligent interface for robot control. The present preliminary study shows the on-going results of the proposed methodology for both self-exploration of robotic morphologies and acquisition of human behaviours.