Maneuvering target tracking is one of the most challenging task in target tracking due to the unpredictable and time-varying nature of target dynamics. The performance of a tracking system depends strongly on how closely the assumed motion models represent the actual motion of the target. A tracking system achieves higher estimation accuracy when the assumed motion model closely matches the target’s motion model (mode). Therefore, the selection and adaptation of motion models play a critical role in accurately estimating the state of maneuvering target. In this talk, we present two approaches: (i) a randomization-based turn-rate computation method and (ii) an Adaptive Transition Probability Matrix (ATPM) approach to provide improved target tracking. These are supported through Monte carlo simulations for different maneuvering target scenarios.

Joint work by: Bibhabasu Mondal, Viji Paul, and Rajbabu Velmurugan