Mobile robot simultaneous localization and mapping based on particle filtering with fixed-lag sampling

Based on the influence analysis of sampling and resampling operations on the consistency of map estimation, two fixed-lag sampling methods are designed to improve the Rao-Blackwellized particle filtering SLAM (RBPF-SLAM) algorithm: the fixed-lag Gibbs sampling and the fix-lag smooth sampling. They both can improve the filterpsilas estimation of past poses and reduce the probability of particle degeneracy. The experiments show that these two sampling strategies can improve the synthetic performance of RBPF-SLAM algorithm: with the same particle number, the times of resampling are decreased and the consistency of map estimation is enhanced remarkably.