Large-Scale UWB Anchor Calibration and One-Shot Localization Using Gaussian Process

TL;DR

This work tackles large-scale UWB calibration and one-shot localization in obstructed environments by fusing CT-LIO with Gaussian Process regression to calibrate anchor positions from UWB ranges using a single sampling pass. By interpolating CT-LIO trajectories to align UWB measurements and employing an iterative GP with a Matérn kernel, the method yields approximately $\sim$2 m anchor accuracy in a $600\times 450$ m$^2$ area, even under NLoS. A modified STD descriptor then leverages calibrated anchors to dramatically reduce search spaces for one-shot localization, improving both accuracy and speed in repetitive logistics environments. The approach is validated on real-world data and is complemented by open-source datasets and calibration code, enabling broader adoption in seaports and warehouses at reduced cost and with greater robustness than traditional UWB-only or LiDAR-Inertial methods.

Abstract

Ultra-wideband (UWB) is gaining popularity with devices like AirTags for precise home item localization but faces significant challenges when scaled to large environments like seaports. The main challenges are calibration and localization in obstructed conditions, which are common in logistics environments. Traditional calibration methods, dependent on line-of-sight (LoS), are slow, costly, and unreliable in seaports and warehouses, making large-scale localization a significant pain point in the industry. To overcome these challenges, we propose a UWB-LiDAR fusion-based calibration and one-shot localization framework. Our method uses Gaussian Processes to estimate anchor position from continuous-time LiDAR Inertial Odometry with sampled UWB ranges. This approach ensures accurate and reliable calibration with just one round of sampling in large-scale areas, I.e., 600x450 square meter. With the LoS issues, UWB-only localization can be problematic, even when anchor positions are known. We demonstrate that by applying a UWB-range filter, the search range for LiDAR loop closure descriptors is significantly reduced, improving both accuracy and speed. This concept can be applied to other loop closure detection methods, enabling cost-effective localization in large-scale warehouses and seaports. It significantly improves precision in challenging environments where UWB-only and LiDAR-Inertial methods fall short, as shown in the video (https://youtu.be/oY8jQKdM7lU). We will open-source our datasets and calibration codes for community use.

Figures (3)

• Figure 1: Survey of UWB-based localization methods and their maximum coverage size, See Sec. II for detailed explanations.
• Figure 2: The system leverages CT-LIO to generate UWB samples for non-parametric GP fitting, calibrating anchor positions. Using these calibrated positions, a range-based search is integrated with the descriptor method for one-shot localization. Additionally, over 100 scans were collected to create a prior map for ground truth verification.
• Figure 3: The proposed solution achieves better prediction accuracy in real-world, large-scale UWB experiments.