A Multi-Scan Labeled Random Finite Set Model for Multi-Object State Estimation
Ba-Ngu Vo, Ba-Tuong Vo · IEEE Transactions on Signal Processing · 2019
State-space models in which the system state is a finite set-called the multi-object state-have generated considerable interest in recent years. Smoothing for state-space models provides better estimation performance than filtering. In multi-object state estimation, the multi-object filtering density can be efficiently propagated forward in time using an analytic recursion known as the generalized labeled multi-Bernoulli (GLMB) recursion. In this paper, we introduce a multi-scan version of the GLMB model to accommodate the multi-object posterior recursion, and develop efficient numerical algorithms for computing this so-called multi-scan GLMB posterior.