Decentralized Cooperative Multi-Robot Localization with EKF

TL;DR

This paper introduces a decentralized multi-robot localization method using EKF that requires minimal landmarks, reduces computational costs, and is robust in featureless environments, validated through simulations and experiments.

Contribution

It presents a novel decentralized EKF-based localization scheme that minimizes storage, reduces landmark requirements, and enhances robustness in cooperative multi-robot systems.

Findings

Effective in featureless environments

Reduces computational and storage costs

Validated by simulations and real experiments

Abstract

Multi-robot localization has been a critical problem for robots performing complex tasks cooperatively. In this paper, we propose a decentralized approach to localize a group of robots in a large featureless environment. The proposed approach only requires that at least one robot remains stationary as a temporary landmark during a certain period of time. The novelty of our approach is threefold: (1) developing a decentralized scheme that each robot calculates their own state and only stores the latest one to reduce storage and computational cost, (2) developing an efficient localization algorithm through the extended Kalman filter (EKF) that only uses observations of relative pose to estimate the robot positions, (3) developing a scheme has less requirements on landmarks and more robustness against insufficient observations. Various simulations and experiments using five robots equipped with relative pose-measurement sensors are performed to validate the superior performance of our approach.