Max Robotics — Global Service Robot Directory + US Certification

About

Lourdes Agapito is a prominent computer vision and robotics researcher whose work spans dense 3D reconstruction, non-rigid tracking, semantic mapping, and visual representation learning for robotic manipulation. She is perhaps best known for Co-Fusion (2017, 221 citations), a landmark contribution to the field of dense SLAM that introduced the ability to simultaneously segment, track, and reconstruct multiple moving objects in real time from RGB-D input — a significant leap beyond traditional systems that assumed static scenes. This work demonstrated both technical ingenuity and practical impact, earning it widespread recognition in the robotics and computer vision communities. Her earlier research on real-time non-rigid structure from motion (2014) laid important groundwork for handling deformable objects in robotic contexts. More recently, Agapito has extended her focus toward robot learning, contributing to self-supervised semantic keypoint discovery for manipulation and few-shot visual imitation through point tracking, as seen in RoboTAP (2024). Her SeMLaPS work further advances real-time semantic mapping for AR/VR and robotic applications. Across her career, Agapito has consistently bridged the gap between geometric 3D understanding and practical robotic intelligence, making her a significant figure in embodied AI research.