Fastslam and Its Data Association Approaches for Mobile Robots

Lijin Guo,Huaxiang Wang,Qinghao Meng,Yanan Qiu
DOI: https://doi.org/10.1109/ical.2007.4338893
2007-01-01
Abstract:This paper discusses the data association issues of the FastSLAM (Factored Solution to SLAM) algorithm that recursively estimates the full posterior distribution of both robot pose and landmark locations. In traditional FastSLAM, it always hypothesizes that the data association is certain. But in real world, the data association is uncertain. In this paper, we present an extension to FastSLAM that handle the uncertainty in the data association using an approach that unites per-particle maximum likelihood data association and negative information technology. Experimental results show that the improved FastSLAM has advantages over the traditional FastSLAM approaches.
What problem does this paper attempt to address?