Ubiquitous social perception abilities for interaction initiation in human-robot interaction

Abstract

Robots acting as assistants or companions in a social environment must be capable of sensing information about the location of the users and analysing and interpreting their social, affective signals in order to be able to plan and generate an appropriate response. Social perception abilities are thereby very important for the robot to evaluate whether it is appropriate or not to initiate an interaction with the user. In this paper we present the initial steps of the design of a ubiquitous social perception system for interaction initiation: users' social signals and expressive behaviour are analysed at different spatial locations and temporal instants. We propose an approach to evaluate whether it is appropriate for a robot to initiate an interaction with the user. We describe an autonomous algorithm to regulate the inter-entity distance between the robot and a person using visual face detection which can be used during interaction initiation and also discuss the role of memory abilities to remember what has happened throughout the interaction.