Markov chain Monte Carlo methods for tracking a maneuvering target in clutter

Abstract

We address the problem of tracking a maneuvering target in a cluttered environment. We propose three algorithms based on stochastic sampling methods to solve the following combinatorial optimization problems: (a) data association, and (b) maneuver detection. The first proposed algorithm is a data augmentation (DA) scheme, that yields conditional mean state estimates of the maneuvering target in clutter. The second proposed scheme is a simulated annealing (SA) version of DA that computes the joint MAP state sequence estimates of the target state, the measurement to target associations and the input maneuvering control sequences. Finally, a SA Metropolis-Hastings DA scheme is designed to yield the MAP state sequence estimate of the measurement to target associations and the input maneuvering control sequences. The cost per iteration is linear in the data length. Furthermore, theoretical convergence results of the three proposed Markov chain Monte Carlo algorithms towards the desired estimates, have been obtained.