A Localization System for Autonomous Vehicles Based on TriLateration Tags

Abstract

An indoor global robot localization problem is considered, where the robot is a differential drive kinematics vehicle. It is equipped with an RFID reader which measures phase shift in the RFID signal back-scattered by a set of RFID passive tags, located on the ceiling of the environment in a known position. The main contribution is the introduction of a special kind of tag, including three antennas close one each other (TriLateration Tag), allowing to obtain an estimate of the robot pose even if only one of these tags is detected. An Extended Kalman Filter is used to fuse odometry readings (encoder steps on robot wheels) with phase measurements obtaining a pose estimate which significantly improves results obtained when using standard one-antenna tags deployed with the same density. Simulation results show the effectiveness of the approach, which is particularly suited for large warehouses where RFID tags cannot be deployed too densely.