A Featureless Approach to Efficient Bathymetric SLAM Using Distributed Particle Mapping

摘要

This paper presents an approach to simultaneous localization and mapping (SLAM) suitable for efficient bathymetric mapping that does not require explicit identification, tracking, or association of seafloor features. This is accomplished using a Rao-Blackwellized particle filter, in which each particle maintains a hypothesis of the current vehicle state and a grid-based, two-dimensional depth map, efficiently stored by exploiting redundancies between different maps. Distributed particle mapping is employed to remove the computational expense of map copying during the resampling process. The proposed approach to bathymetric SLAM is validated using multibeam sonar data collected by an autonomous underwater vehicle over a small-timescale mission (2 h) and a remotely operated vehicle over a large-timescale mission (11 h). The results demonstrate how observations of the seafloor structure improve the estimated trajectory and resulting map when compared to dead reckoning fused with ultrashort-baseline or long-baseline observations. The consistency and robustness of this approach to common errors in navigation is also explored. Furthermore, results are compared with a preexisting state-of-the art bathymetric SLAM technique, confirming that similar results can be achieved at a fraction of the computation cost.