Engineering statistics for multi-object tracking

Progress in single-sensor, single-object tracking has been greatly facilitated by the existence of a systematic, rigorous, and yet practical engineering statistics that supports the development of new concepts. Surprisingly, until recently no similar engineering statistics has been available for multi-sensor, multi-object tracking. I describe the Bayes filtering equations (the theoretical basis for all optimal single-sensor, single-object tracking) and explain why their generalization to multisensor-multitarget problems requires systematic engineering statistics - i.e., finite-set statistics (FISST). I conclude by summarizing the main concepts of FISST - in particular, the multisensor-multitarget differential and integral calculus that is its core.