V2I Work Zone Geometry Reconstruction with Pose-Conditioned UWB Range Denoising
Researchers have developed a new method for reconstructing work zone geometry using ultra-wideband (UWB) range data from connected and autonomous vehicles (CAVs). This approach utilizes a pose-conditioned, permutation-equivariant predictive denoiser to improve the accuracy of UWB range estimations, which are often degraded by outliers and non-line-of-sight errors. The system incorporates vehicle motion as a geometric prior and was evaluated on real-world field data, showing a significant reduction in measurement-weighted mean squared error. AI
IMPACT Enhances safety and efficiency for autonomous vehicles navigating complex road environments.