Nathan Piasco

Huawei Technologies (France), Centre National de la Recherche Scientifique

Papers

Total Citations

H-Index

About

Nathan Piasco is a researcher at the forefront of visual localization for robotics and autonomous vehicles, specializing in making machines reliably understand where they are from camera images alone. His core research spans camera pose regression, uncertainty estimation, and the innovative use of neural scene representations for localization. Piasco’s most influential work, “CoordiNet,” introduced a CNN-based algorithm that directly predicts 3D position and orientation from a single image while also providing a measure of prediction uncertainty—a critical capability for safe autonomous navigation. This paper has garnered 41 citations, reflecting its impact on reliable vehicle localization. In “LENS,” he pioneered the use of Neural Radiance Fields (NeRF) to generate synthetic training data, demonstrating that novel view synthesis can significantly improve camera pose regression under challenging conditions. Complementing this, his work on fusing auxiliary modalities further enhanced localization robustness. Through these contributions, Piasco is advancing the practical deployment of vision-based localization systems, bridging the gap between cutting-edge neural rendering and real-world robotic perception.