Performance estimation for Kalman filter based multi-agent cooperative navigation by employing graph theory

Abstract

Cooperative Navigation (CN) exploits inter-agent relative measurement and communication to achieve navigation performance improvement. This technique has attracted worldwide interest in many multi-agent (e.g., multiple unmanned air vehicles) applications, due to its significant advantage over non-cooperative approaches. Comparing to prior studies which mainly focused on performance analysis under a fixed CN framework, this work aims at establishing the theoretical basis to quantify the navigation performance for different CN integration architectures. Two graph variables are defined to describe the architecture: relative measurement graph and communication & fusion graph. The measurements are integrated through extended Kalman filters, and the state covariance matrices are rigorously derived and bounded to establish the relationship between the navigation performance and the integration architecture. Simulations are carried out to demonstrate and validate the proposed framework, and the results show its feasibility and effectiveness.