Abstract: In this paper, we present the design and performance analysis of an innovative system for tracking Automated Guided Vehicles (AGVs) in indoor industrial scenarios. An on-board odometer provides information about the dynamic state of the AGV, allowing to predict its pose (i.e., its position and orientation). At the same time, an external Ultra-Wide Band (UWB) wireless network provides the information necessary to compensate for the error drift accumulated by the odometer. Two novel alternative solutions for AGV tracking are proposed: (i) a classical Time Differences Of Arrivals (TDOA) approach with a single receiver; (ii) a "Twin-receiver" TDOA (TTDOA) approach, that requires the presence of two independent receivers on the AGV. The TTDOA configuration allows to indirectly estimate the orientation of the vehicle, thus increasing the estimation accuracy. Moreover, this allows direct estimation of the vehicle's movement even when the odometer is not working properly (e.g., temporary failure) or when the AGV is not moving (e.g., at the start-up). The system performance with the two proposed tracking algorithms is evaluated in realistic conditions, by considering a consolidated UWB channel model and a simple on-board energy detector receiver. The impact of the wireless network architecture and of the presence of moving obstacles is analyzed. The obtained results show clearly that the implementation of a tracking system with a sub-centimeter accuracy can be realized by means of low-complexity UWB receiver and commercial odometers. The automatic movement of goods within a warehouse is one of the most appealing application of the proposed tracking system.
Keywords: Ultra wideband, UWB, automated guided vehicle, AGV, localization, tracking, time difference of arrivals, TDOA
