Real-Time Initialization of Unknown Anchors for UWB-aided Navigation

TL;DR

The paper tackles automatic real-time initialization of unknown UWB anchors in GNSS-denied navigation by introducing a PDOP-based trigger, online outlier rejection, and an adaptive robust nonlinear optimization pipeline for tag-to-anchor measurements. It proves that a closest-point-to-anchor PDOP estimate is conservatively biased and does not require prior anchor positions, enabling reliable and fast anchor calibration during operation. Extensive simulations and real-world AMR and UAV experiments demonstrate improved initialization robustness, faster convergence, and reduced positioning error compared with state-of-the-art methods, aided by an open-source C++ library with a ROS wrapper. The work significantly enhances practical deployability of UWB-aided navigation in complex environments.

Abstract

This paper presents a framework for the real-time initialization of unknown Ultra-Wideband (UWB) anchors in UWB-aided navigation systems. The method is designed for localization solutions where UWB modules act as supplementary sensors. Our approach enables the automatic detection and calibration of previously unknown anchors during operation, removing the need for manual setup. By combining an online Positional Dilution of Precision (PDOP) estimation, a lightweight outlier detection method, and an adaptive robust kernel for non-linear optimization, our approach significantly improves robustness and suitability for real-world applications compared to state-of-the-art. In particular, we show that our metric which triggers an initialization decision is more conservative than current ones commonly based on initial linear or non-linear initialization guesses. This allows for better initialization geometry and subsequently lower initialization errors. We demonstrate the proposed approach on two different mobile robots: an autonomous forklift and a quadcopter equipped with a UWB-aided Visual-Inertial Odometry (VIO) framework. The results highlight the effectiveness of the proposed method with robust initialization and low positioning error. We open-source our code in a C++ library including a ROS wrapper.

Figures (11)

• Figure 1: The two different mobile robots used in the real-world demonstration of our anchor initialization method: an automated forklift (left) and a quadcopter (right). The forklift uses two UWB receivers for 3D anchor positioning, as a single tag cannot provide vertical axis information for a ground vehicle. A single UWB receiver is sufficient for the quadcopter, assuming movement along the vertical axis.
• Figure 2: Diagram illustrating the proposed framework for initializing unknown UWB anchors within a UWB-aided VIO system. A key feature is the real-time PDOP estimation, which triggers anchor initialization once it drops below a defined threshold, preventing poor initialization due to unfavorable geometric configurations. This approach allows for the continuous detection and initialization of new anchors becoming available during operation.
• Figure 3: Outlier rejection method using an online consistency check. The absolute differences between consecutive range measurements $\Delta d$ and tag positions $\Delta p$ are compared. Outliers are identified and rejected when the condition ${\Delta d < \Delta p + \tau}$ is not met, with $\tau$ being an adjustable threshold.
• Figure 4: Outlier rejection performance in a real-world experiment with an automated forklift. This figure shows the effectiveness of the proposed online outlier rejection method applied to real-world UWB range data. Red dashed lines indicate outliers that fall outside the plot range. The green box provides a sampled zoomed-in view, highlighting that smaller outliers are also detected.
• Figure 5: Geometric representation of initialization regions and their influence on our closest-point-to-anchor PDOP estimation from \ref{['equ:gtilde']}. When initialization occurs in the green zone, which is the typical case, e.g., when anchors are attached to walls, the PDOP estimation remains conservative. If initialization takes place in the blue zone, it means that the trajectory "wraps around" the anchor, and thus, the PDOP is inherently low due to good geometric condition, rendering our then slight overconfidence in the metric negligible.

Theorems & Definitions (2)

• Theorem 3.1
• proof