Two measurement scenarios for anonymous mutual localization in multi-UAV systems

We present a method for reconstructing the relative poses among the components of a multi-UAV system using anonymous (i.e., without identity information) robot-to-robot measurements. We consider two cases: bearing-only and bearing+distance measurements. While bearing can be rather directly extracted from a camera image, visual reconstruction of distances is more elaborate and typically associated with a larger noise. Nevertheless, our experiments show that use of such metric information improves signicantly the quality of the localization.