Self Contained Relative Localization with a Low-Cost Multi-Robot System

TL;DR

This paper introduces a centralized, low-cost method for relative localization of multi-robot systems in GPS-denied environments, utilizing sensors, beacons, and a UKF algorithm, achieving robot-sized accuracy.

Contribution

It presents a novel, cost-effective hardware setup and a centralized UKF-based algorithm for relative localization in multi-robot systems without GPS.

Findings

Localization errors are comparable to robot size.

The method works in environments lacking GPS or motion capture.

Hardware and algorithm are cost-effective and practical.

Abstract

A key limitation of current multi-robot systems is a lack of relative localization, particularly in environments without GPS or motion capture systems. This article presents a centralized method for relatively localizing a 2D swarm using sensors and beacons on the robots themselves. The UKF-based algorithm as well as the requisite novel and cost-effective sensing hardware are discussed. Comparisons with a motion capture system show that the method is capable of localization with errors on the order of the size of the robots.