The paper introduces GAIA, a geometry-aware, infrastructure-anchored learning framework for denoising outdoor ultra-wideband ranges used in work-zone reconstruction. It combines temporal range modeling, latent anchor-layout estimation, and deterministic distance projection while retaining range denoising as the supervised task. The authors evaluate it on a real-world dataset with synchronized UWB, GNSS, and IMU measurements, plus a real-data-calibrated stress-test simulator. Against filtering-based and learning-based baselines, GAIA reportedly achieves the lowest range MSE and highest polygon IoU, with an 18.4% MSE reduction and 15.5% IoU improvement over PoseMLP.
No heat snapshots are available in the last 24 hours.