Distributed strong tracking unscented particle filter for simultaneous localization and mapping

The simultaneous localization and mapping (SLAM) based on a conventional centralized filter reconfigures the entire state vectors in every necessary cycle as the number of landmarks changes, which is result in an exponential growth in computation quantities and hard to isolate potential faults. For that, SLAM system using distributed particle filter was presented to cope with these problems. In this paper, the distributed strong tracking unscented particle filter (DSTUPF) is presented to improve the idea of SLAM system based on distributed particle filter. The unscented particle filter (UPF) was used in every local filter to increase the estimation performance and the configuration of proposed system was introduced. However, UPF lacks ability of adaptive adjustment on-line. To deal with this problem, this paper proposes an improved SLAM algorithm that combines the strong tracking filter (STF) and UPF, STF has good performance for adjusting the filter gains on-line, it satisfies the demand of algorithm which has self-adapted ability. The experiment results show that the DSTUPF-SLAM reduces computation quantities compared to the centralized particle filter and is capable of improving estimation performance.