Distributed target localization using a group of UGVs under dynamically changing interaction topologies

Abstract

This paper presents a distributed Bayesian filtering (DBF) method for a network of multiple unmanned ground vehicles (UGVs) under dynamically changing interaction topologies. The information exchange among UGVs relies on a measurement dissemination scheme, called Latest-In-and-Full-Out (LIFO) protocol. Different from statistics dissemination approaches that transmit posterior distributions or likelihood functions, each UGV under LIFO only sends a buffer that contains latest available measurements to neighboring nodes, which significantly reduces the transmission burden between each pair of UGVs to scale linearly with the size of the network. Under the condition that the union of undirected switching topologies is connected frequently enough, LIFO can disseminate observations over the network within finite time. The LIFO-based DBF algorithm is then derived to estimate individual probability density function (PDF) for target localization in a static environment. The consistency of this algorithm is proved that each individual estimate of target position converges in probability to the true target position. The effectiveness of this method is demonstrated by comparing with consensus-based distributed filters and the centralized filter in simulations.