Berta Bescos

Papers

Total Citations

945

H-Index

About

Berta Bescos is a pioneering researcher in computer vision and robotics, specializing in visual SLAM (Simultaneous Localization and Mapping), dynamic scene understanding, and privacy-preserving localization. She is best known for developing DynaSLAM, a groundbreaking system that overcame one of the most fundamental limitations in SLAM research — the rigid-scene assumption — by enabling robust tracking, mapping, and inpainting in dynamic real-world environments populated by moving objects. Published in 2018, this work has amassed an remarkable 924 citations, reflecting its transformative impact on the robotics and autonomous systems communities. Building on this foundation, Bescos extended her contributions with DynaSLAM II, introducing tightly-coupled multi-object tracking tailored for applications such as autonomous driving and augmented reality. Her research portfolio also demonstrates a forward-thinking concern for ethical AI deployment; her work on privacy-preserving inpainting addresses the sensitive challenge of protecting personal biometric data within cloud-based visual localization services. Across her career, Bescos has consistently pushed the boundaries of what mobile robots and autonomous systems can achieve in complex, unstructured environments, making her a significant voice in modern robotics research.