Adaptive iterated particle filter

Abstract

The adaptive iterated particle filter (AIPF) is presented, where the importance density function is updated iteratively by the particle filter itself when necessary. By using a simulated annealing algorithm with an adaptive annealing parameter, the current measurement can be quickly incorporated into the sampling process, resulting in greatly improved sampling efficiency. Simulation results demonstrate the improved performance of the AIPF over the sampling importance resampling filter, unscented Kalman particle filter and auxiliary particle filter. © The Institution of Engineering and Technology 2013.