"A Self-Contained Distributed Sensing System for Swarm Robotics"
Department: Electrical and Computer Engineering
Advisor: Dr. Jon Wallace
Abstract:
Swarm robotics is a field currently experiencing great interest from the research community. The key idea is that a large number of simple robots working together can accomplish certain tasks more effectively than a traditional expensive, complex robot. Swarms are particularly useful in applications benefiting from a swarm’s characteristics of parallelization and redundancy. One such application is the mapping or exploration of hostile environments. However, currently there is minimal real-world utilization of swarms outside of research labs. Part of this problem has to do with robot localization in most of the literature being performed by external motion capture systems. Furthermore, sensing systems tend to be fairly expensive. These systems require experiments to be performed in controlled lab environments, and individual robot costs can be prohibitive for swarms with many members. In this presentation, a low-cost hardware platform for swarm robotics will be presented. The platform is capable of globally localizing the swarm by synthesizing local measurements taken by sensors on board the robots without the need for external sensors. In this way, a picture of robot locations similar to that which an external sensing system could generate can be made using sensors which are entirely self-contained on the robots themselves. Care has also been taken to minimize the individual robot hardware cost as much as possible such that the platform is accessible for use in other labs and many robots can be built relatively affordably. Control software and a simulation environment for testing have also been developed. The hardware design, sensing system, and control algorithm will be discussed. Limitations and possible directions for future research will be briefly presented.
