Estimation and control of multi-object systems with high-fidenlity sensor models: A labelled random finite set approach

Abstract

Principled and novel multi-object tracking algorithms are proposed, that have the ability to optimally process realistic sensor data, by accommodating complex observational phenomena such as merged measurements and extended targets. Additionally, a sensor control scheme based on a tractable, information theoretic objective is proposed, the goal of which is to optimise tracking performance in multi-object scenarios. The concept of labelled random finite sets is adopted in the development of these new techniques.