This talk will outline the progress of research, and the state of the art, in the random finite set (RFS) or finite set statistics (FISST) paradigm, of multi-object estimation or multi-target tracking. For characterizing systems with stochastically time-varying numbers of targets or states, which are observed in the presence of noise, clutter and missed detections, the RFS or FISST paradigm has attracted substantial interest from both the academic and industrial community. We review the well known PHD filter which is most commonly associated with the RFS/FISST family and discuss its strengths and limitations. We discuss more recent works on Cardinalized PHD or CPHD filtering, as well as Multi-Bernoulli or MB filtering. Finally we discuss the state of the art in the form of a conjugate prior for the multi-object posterior. This approach also represents a Bayes optimal approach to track management. The end result establishes a closed form recursion, for the time prediction and data update steps of the Bayes multi-object filter, and gives rises to tractable implementations and efficient approximations. Applications are presented throughout the discussion.

Biography:Ba Tuong Vo is currently an Associate Professor with the Department of Electrical and Computer Engineering at Curtin University in Perth, Australia. He obtained his Bachelor degrees in Science majoring in Applied Mathematics, Engineering majoring in Electrical and Electronic Engineering, and his PhD with distinction all from The University of Western Australia. He has active research interests in multi-target tracking, Bayesian estimation, and statistical signal processing.