Using Group Knowledge for Multitarget Terrain-Based State Estimation

Abstract

Multitarget terrain-based tracking is a cyclic process that combines sensor information with state estimation and data association techniques to maintain an estimate of the state of an environment in which ground-based vehicles are operating. When the ground-based vehicles are military vehicles moving across terrain, most of them will being moving in groups instead of autonomously. This work presents a methodology that has been demonstrated to improve the estimation aspect of the tracking process for this military domain. A clustering algorithm identifies groups within a vehicular data set. Group characteristics are extracted and then a new, innovative technique is utilized to integrate these into the individual vehicles’ state estimation process. A series of experiments shows that the proposed methodology significantly improves the performance of three classic estimation algorithms for multitarget terrain-based tracking.