PF-UKF-RJMCMC Approaches for Radar Target-Tracking

Real-time agent-based crowd simulation with the Reversible Jump Unscented Kalman Filter

“…Markov Chain Monte Carlo (MCMC) techniques [36,37,38] have been applied to this problem with success being seen in Reversible Jump MCMC (RJM-CMC) [11]. RJMCMC performs well in sampling both the choice of model and its associated parameters of different dimensions simultaneously using a modified Bayes factor.…”

“…any new information requires re-calibrating for all previous data. Attempts at sequential applications of RJMCMC have been made [38] but these techniques are not efficient enough for real time application due to the high dimensionalily of ABMs. This paper proposes to combine RJMCMC with the UKF to produce an efficient algorithm that can apply multiple model inference to ABMs in real time.…”