Effective joint probabilistic data association using maximum a posteriori estimates of target states

In multi-target tracking (MTT), closely moving targets can be tracked accurately with access to the true measurement-to-track (M-to-T) association configuration, which is not known in practical scenarios. The joint probabilistic data association approach (JPDA) computes the association configuration probabilities using the predicted target states. In MTT when targets are close, the target states are obtained by marginalizing the target states computed over all feasible association configurations. The proposed approach computes the association configuration probabilities using estimated posterior target states and improves the estimation accuracy. The posterior target states are obtained as maximum a posteriori (MAP) estimates. The MAP estimates are obtained using an iterative procedure based on the expectation maximization algorithm. Compared to other data association approaches such as Set JPDA and JPDA* the proposed approach provides better estimation accuracy and improves track identity maintenance for closely moving targets. The identity is maintained in the proposed approach by using the estimated posterior target states for computation of association configuration probabilities. Monte Carlo simulations verify the advantage of the proposed method over other approaches in a simulated, multi-target, cluttered environment with varying probabilities of detection.