RFS Collaborative Multivehicle SLAM: SLAM in Dynamic High-Clutter Environments

Recently, we proposed a novel solution to the collaborative multivehicle simultaneous localization and mapping (CMSLAM) problem by extending the random finite set (RFS) SLAM filter framework using recently developed multisensor information fusion techniques in the finite set statistics. We modeled the measurements and the landmark map as RFSs, and a joint posterior consisting of the landmark map and the vehicle trajectories was propagated in time to solve the CMSLAM problem. The proposed solution is based on the Rao?Blackwellized particle filter-based vehicle trajectories posterior estimation and the probability hypothesis density (PHD) filter-based landmark map posterior estimation. In this article, we evaluate the performance of this solution under dynamic high-clutter environmental conditions using a series of simulations and an actual experiment.