On the value of collaboration in anchorless robot self-localization

In this paper, we consider the value of collaboration in anchorless robot self-localization. Robots which rely solely on data from imperfect inertial measurement units become increasingly less certain of their locations over time. Without anchors or landmarks to help them regain their bearings, these robots reach a point at which they essentially have no idea of their locations. If, however, these robots are given the ability to communicate with one another, they can utilize additional information gained through collaboration to correct their beliefs and collectively improve their location estimates. In this work, we characterize how inter-robot collaboration impacts self-localization performance and discuss whether any additional information, such as an estimator confidence measure, may be gleaned from the resulting corrected beliefs.