Ultra-wideband Time Difference of Arrival Indoor Localization: From Sensor Placement to System Evaluation

TL;DR

This work addresses the gap in practical indoor localization by treating sensor placement as a core system-level design variable in UWB TDOA localization and validating the approach through real-world deployment. It combines sensor-placement optimization with a UWB TDOA–IMU fusion via an error-state Kalman filter in a 16‑D state, using a BCM-based optimization to minimize the average RMSE bound over target points. Across multi-room, staircase, and occlusion scenarios, the system achieves sub-decimeter to tens-of-centimeter accuracy (as low as ~16–28 cm RMSE) while maintaining low cost and portability, and the results are closely compared to theoretical lower bounds to quantify placement impact. The study provides practical deployment guidelines, demonstrates robustness to occlusion and challenging geometries, and commits to releasing datasets to facilitate further research and benchmarking in UWB-based indoor localization.

Abstract

Wireless indoor localization has attracted significant research interest due to its high accuracy, low cost, lightweight design, and low power consumption. Specifically, ultra-wideband (UWB) time difference of arrival (TDOA)-based localization has emerged as a scalable positioning solution for mobile robots, consumer electronics, and wearable devices, featuring good accuracy and reliability. While UWB TDOA-based localization systems rely on the deployment of UWB radio sensors as positioning landmarks, existing works often assume these placements are predetermined or study the sensor placement problem alone without evaluating it in practical scenarios. In this article, we bridge this gap by approaching the UWB TDOA localization from a system-level perspective, integrating sensor placement as a key component and conducting practical evaluation in real-world scenarios. Through extensive real-world experiments, we demonstrate the accuracy and robustness of our localization system, comparing its performance to the theoretical lower bounds. Using a challenging multi-room environment as a case study, we illustrate the full system construction process, from sensor placement optimization to real-world deployment. Our evaluation, comprising a cumulative total of 39 minutes of real-world experiments involving up to five agents and covering 2608 meters across four distinct scenarios, provides valuable insights and guidelines for constructing UWB TDOA localization systems.

Figures (10)

• Figure 1: A conceptual diagram demonstrating the deployment of an indoor localization system in a shopping mall. The image illustrates heterogeneous agents, including ground robots and flying robots, leveraging the indoor positioning system to navigate seamlessly alongside customers while providing various services, such as item delivery or cleaning service. (This conceptual image is created from an image we generated using DALL·E 3 text-to-image models developed by Open AI.)
• Figure 2: Conceptual diagrams for UWB TWR (left) and TDOA (right) localization system. In TWR, the UWB tag actively communicates with UWB anchors for localization. In TDOA, the UWB tag listens to the communications between anchors passively for positioning.
• Figure 3: The system diagram provides a comprehensive overview of each component in our UWB TDOA localization system. The experimental setup of the multi-agent pedestrian localization is shown in (b). UWB anchors, enclosed by red circles, are installed in the space with positions surveyed by a Leica total station (a). The heatmap in (c) illustrates the localization performance of the anchor constellation shown in (b), calculated at a height of $1.5$ meters using sensor placement analysis. Lower root-mean-squared error (RMSE) is indicated as darker color. The hardware components of our UWB handheld device along with the onboard ESKF localization algorithm are shown in (d).
• Figure 4: The multi-agent pedestrian localization experiment with five agents in constellation $1$ is shown in the photo above. The estimated together with the ground truth trajectories of each agent during the experiments are shown in the bottom plots. Readers are encouraged to view our supplementary video (http://tiny.cc/uwb_tdoa_ram24) to gain a better insight into our experimental process and evaluate the robustness of the localization performance.
• Figure 5: Comparison of the average experimental root-mean-square error (RMSE) for multi-agent pedestrian localization in Const. $1$ and $2$, along with localization under occlusion in Const. $1$ (Const. 1 Backpack), shown as blue bars, against the theoretical RMSE lower bounds represented by green bars.