École Nationale Supérieure de
Techniques Avancées

Abstract

Social robots involved in Human-Robot Interaction (HRI) require highly accurate user detection. To enhance the process, fusion of data coming from several sensors or detectors can be resorted to. The coming work offers a brief review of user detection and data fusion techniques. The goal is to provide a first first framework for further enhancements of the use of fusion, based on the fusion process developed for the recent STRANDS project. The study first focuses on user detection to give an evaluation of the performances of various detectors. Then, a solution to integrate the STRANDS process is submitted. Several scenarios are investigated to assess the performances and advantages of the data fusion.