Home /Research /Visual recognition of conspecifics by swarm robots
SWARM

Visual recognition of conspecifics by swarm robots

Robin Andrew Russell

Year
2004
Citations
2

Abstract

This paper examines the problem of giving very simple swarm robots the capability of recognizing and locating other swarm members. The ability to recognize conspecific robots will allow swarms of robots to adjust their mutual separation as required by any specific task. For instance, robots engaging in flocking behavior will need to maintain a fixed mutual separation. One of the perceived benefits of developing swarm robots is that the individual robots will be simpler than a single complex robot built for the same task. With this in mind it is essential that the robot recognition system be as simple as possible, otherwise part of the justification for using swarms is negated. For this project a visual recognition scheme based on considering the abilities of social insects has been developed. Hardware and processing capabilities are minimal making the scheme suitable for all but the simplest robot systems. In this paper details are given of the camera hardware, and imageprocessing steps required to determine the range and identity of robots carrying visual markings. Results of practical robot recognition experiments are also presented. 1

Keywords

RobotSwarm behaviourComputer scienceSwarm roboticsArtificial intelligenceComputer vision

Related papers

Browse all SWARM papers