Impact of Simultaneous Localization and Calibration on Closed-Loop Cooperative Robot Radio Navigation

Abstract

Control of a team of mobile robots requires accurate pose estimates, which is particularly challenging when globally referenced external localization systems, like GNSS, are not available. Cooperative radio navigation, where robots exchange radio signals with each other and with fixed devices known as anchors, can provide accurate and robust localization. However, in real implementations accurate calibration is challenging and effects such as temperature variations can decalibrate radio transceivers, introducing ranging biases that degrade pose estimates and, consequently, closed-loop control performance. In this work, we evaluate how biased radio ranging measurements affect trajectory tracking in a closed-loop control system, and how online calibration of these biases improves the overall navigation performance. To this end, a centralized EKF-based simultaneous localization and calibration (SLAC) framework is used to jointly estimate robot poses and node-wise ranging biases, and the resulting pose estimates are provided as feedback to a trajectory tracking controller. The SLAC-based feedback is compared with cooperative localization without bias calibration. Monte Carlo simulations show that SLAC improves both position estimation and tracking accuracy by approximately one order of magnitude compared with cooperative localization without ranging bias calibration.