Distributed Joint Mapping and Registration with Limited Fields-of-View

摘要

The interest of this paper is to fuse maps, expressed in different local coordinates, of multiple mobile agents with limited fields-of-view, without knowledge of the relative poses among agents. To this end, the map is modeled as a multi-object Poisson process and a recently developed approach, named Cauchy-Schwarz fusion (CSF), is adopted. In order to perform calibration among local coordinates, the information-theoretic interpretation of CSF is exploited to build a suitable map discrepancy to be minimized with respect to the drift and orientation parameters. A multi-hypothesis method is adopted in order to efficiently solve the resulting optimization problem.