Recent advances in multi-INT track fusion

This paper provides recent experimental results on a large-scale scenario for recently-introduced MCMC-based and MHT-based solutions to the multi-INT problem, where sparse non-kinematic data is to be fused with multi-sensor kinematic data. The MCMC Data Fuser (MCMC-DF) and Asynchronous MHT (A-MHT) show promising scalability relative to classical track-oriented MHT, the leading paradigm for high-performance multi-target tracking. Additionally, we propose a number of modeling and algorithmic enhancements: (i) efficient hypothesis formation based on coarse (pairwise) track gating; (ii) a forward-backward Kalman smoother for Ornstein-Uhlenbeck (OU) processes; (iii) an OU-IMM filter for the target move-stop-move scenarios that are of interest; and (iv) fusion logic and scoring generalizations to accommodate multiple emitters per target.