Bertrand Le Saux

Université Paris-Saclay, Office National d'Études et de Recherches Aérospatiales

Papers

Total Citations

H-Index

About

Bertrand Le Saux is a leading researcher in computer vision and robotics, with a focus on 3D scene understanding and multimodal perception. His most influential contribution is the development of **SnapNet-R**, a pioneering framework for consistent 3D multi-view semantic labeling that enables robots to semantically interpret their environments from reconstructed 3D data. This work, which has garnered **82 citations**, addresses a critical challenge in robotics: how to maintain coherent semantic labels across multiple viewpoints as a robot navigates and reconstructs its surroundings. Le Saux's approach synthesizes 3D-coherent scene observations, allowing for robust object and environment recognition directly from sensor data or motion-derived 3D models. Additionally, his research on **modality-independent classifiers** for people detection explores innovative fusion techniques for RGB and depth information, particularly in robotic discovery scenarios. This work demonstrates his commitment to making robots more perceptive and interactive in human environments. Through his contributions to semantic mapping and multimodal fusion, Le Saux has advanced the frontier of autonomous robotic perception, enabling machines to understand and navigate complex, unstructured spaces with greater accuracy and reliability.