Socially acceptable approach to humans by a mobile robot

Abstract

In this paper we propose a solution to endow a mobile robot with the ability to approach humans in a safe and socially acceptable way. Our proposal focuses on real world indoor environments where the usual presence of multiple humans and obstacles notably rise the complexity of the approach action. We first deal with the problem of accurately estimating the 3D poses of all humans in the work space (positions and orientations), to then focus on the estimation of the most appropriate navigation goal from which the robot should start the interaction with the user. For the latter, we define a cost function that, accounting for multiple proxemic parameters and complying with the restrictions inherent of a robot navigating in a human environment, enables ascertaining an optimal solution. Different experiments are presented to demonstrate the feasibility of our proposal to work under real world conditions.