A Loosely Coupled Cooperative Localization Augmentation to Improve Human Geolocation in Indoor Environments

Abstract

This paper reports on the use of a cooperative localization augmentation to increase the localization accuracy of human agents in an opportunistic fashion by processing interagent relative measurements. The main challenge in the decentralized cooperative localization algorithm design is how to account for the strong correlations, which the relative measurement updates create between the state estimates of the agents, with a reasonable communication cost. To keep track of the correlations agents need to communicate with each other through some form of a network-wide communication topology, which is hard to maintain for human agent localization applications. In this paper, we discuss a cooperative localization method that, instead of maintaining the correlations, accounts for them in an implicit manner by using conservative upper-bound estimates on the joint correlation matrix of the agents. This provably consistent loosely coupled cooperative localization method requires only the two agents involved in a relative range measurement to communicate with each other. Our results include the use of this algorithm for human agent localization via UWB ranging sensors. We demonstrate our results in simulation and experiments.