Decentralized Cooperative Localization: A Communication-Efficient Dual-Fusion Consistent Approach

Abstract

Decentralized cooperative localization poses significant challenges in managing inter-robot correlations, especially in environments with limited communication capacity and unreliable network connectivity. In this letter, we propose a communication-efficient decentralized consistent cooperative localization approach with almost minimal requirements for storage, communication, and network connectivity. A dual-fusion framework that integrates heterogeneous and homogeneous fusion is presented. In this framework, each robot only tracks its own local state and exchanges local estimates with its neighboring robots that possess relative measurements. In the heterogeneous fusion stage, we present an MAP-based decentralized fusion approach to fuse prior estimates of multiple heterogeneous states received from neighboring observed robots and nonlinear measurements in the presence of unknown cross-correlations. In the homogeneous fusion stage, the estimates from neighboring observing robots are further fused based on the CI technique, fully exploiting all available information and thus yielding better estimation results. The proposed algorithm is proved to be consistent. Extensive Monte Carlo simulations and real-world experiments demonstrate that our approach outperforms state-of-the-art methods.