A PMHT Approach for Extended Objects and Object Groups

Abstract

Conventional tracking algorithms rely on the assumption that the targets of interest are point source objects. However, in realistic scenarios the point source assumption is often not suitable and estimating the object extent becomes a crucial aspect. Recently, a Bayesian approach to extended object tracking using random matrices has been proposed. Within this approach, ellipsoidal object extensions are modeled by random matrices and treated as additional state variables to be estimated. However, only a single-object solution has been presented so far. In this work we present the multi-object extent of this approach. We derive a new variant of probabilistic multi-hypothesis tracking (PMHT) that simultaneously estimates the ellipsoidal shape and the kinematics of each object using expectation-maximization (EM). Both the ellipsoids and the kinematic states are iteratively optimized by specific Kalman filter formulae that arise directly from the PMHT framework. The novel method is demonstrated and evaluated by simulations.