In this work, we propose a new simultaneous localization and mapping (SLAM) approach that allows to learn the floor plan representation and to deal with inaccurate information. A key feature is an online estimated channel characterization that enables an efficient combination of the measurements. Starting with just the known anchor positions, the proposed method includes the virtual anchor (VA) positions also in the state space and is thus able to adapt the VA positions during tracking of the agent. Furthermore, the method is able to discover new potential VAs in a feature-based manner. 

The work presents a proof of concept using measurement data. The excellent agent tracking performance of  90 % of the error lower than 5 cm achieved with a known floor plan can be reproduced with SLAM.