VISOR Proposal

Visual Perception System for a Social Robot

Proposer: Francisco Sandoval Hernández, ETSI Telecomunicacion, Universidad de Málaga, Spain;

Accepted: 2005. Final Report.

Abstract

The aim of this research study is to generate in-depth knowledge of the visual human-oriented perception system for a socially interactive robot: robot for which social human-robot interaction is an important task. The human-oriented perception system is one of the issues imposed by social interaction that a social robot must address. If most human-oriented perception is based on passive sensing (artificial vision and auditory), the vision system is the responsible of solve the problems of identifying faces, measuring head and hands poses, capturing human motion, recognizing gestures and reading facial expressions to emulate human social perception. This information permit that the robot be able to identify who the human is, what the human is doing, how the human is doing it and even to imitate the human motion. Thus, the robot could treat the human as an individual, understand his/her surface behaviour, and potentially infer something about his/her internal states (e.g., the intent or the emotive state). On the other hand, these human-related tasks must be run in parallel with object-related ones, which permit the robot to recognize objects extracted from the environment. Our research study will investigate how the requirements for accomplishing a visual task determine the optimal architecture of a vision system. The issues that will be considered are: human and object representation, selection of low-level and high-level vision modules and control schemata. Traditionally, computer vision research has emphasized investigating each vision module as a general and isolated item. Such research efforts often generate unrealistic solutions from ill-defined assumptions. In contrast, we will consider a vision system as a whole and emphasize the research on interactions among modules as well as the research on each vision module.

Consortium

• Francisco Sandoval Hernández, ETSI Telecomunicacion, Universidad de Málaga, Spain, (Project co-ordinator),
• Jorge Dias, ISR, Universidade de Coimbra, Portugal.