Francesc Moreno-Noguer

About

Francesc Moreno-Noguer is a prominent researcher whose work spans computer vision, robotics, and human-robot interaction, with particular expertise in 3D scene understanding, robotic manipulation, and human pose estimation. His research has made significant strides in enabling machines to perceive and interact with the physical world more intelligently. Among his most influential contributions is *GanHand* (2020, 170 citations), a pioneering framework for predicting human grasp affordances in multi-object scenes from single RGB images — a problem previously unexplored. His early work on robotic cloth manipulation, including depth-aware grasping strategies (2012, 131 citations) and the fast 3D textile descriptor FINDDD (2013), established foundational tools for handling deformable materials in robotic contexts. Moreno-Noguer has also advanced non-rigid structure-from-motion using physical priors and uncertainty-aware camera pose estimation, contributing to robust localization in AR/VR and robotics systems. More recently, his *PoseFix* work (2023) explores natural language-guided 3D human pose correction, opening avenues for personalized coaching and physical therapy applications. Through over a decade of interdisciplinary research, Moreno-Noguer has consistently bridged perception and action, making him a leading voice in intelligent robotic systems and visual understanding.

Research Focus

Key Achievements

Top Papers

1. 1
   GanHand: Predicting Human Grasp Affordances in Multi-Object Scenes

   170 citations · 2020
2. 2
   Using depth and appearance features for informed robot grasping of highly wrinkled clothes

   131 citations · 2012
3. 3
   Sequential Non-Rigid Structure from Motion Using Physical Priors

   80 citations · 2015
4. 4
   Learning RGB-D descriptors of garment parts for informed robot grasping

   53 citations · 2014
5. 5
   Teaching Robot’s Proactive Behavior Using Human Assistance

   42 citations · 2017
6. 6
   Uncertainty-Aware Camera Pose Estimation from Points and Lines

   36 citations · 2021
7. 7
   FINDDD: A fast 3D descriptor to characterize textiles for robot manipulation

   35 citations · 2013
8. 8
   Discrete time motion model for guiding people in urban areas using multiple robots

   34 citations · 2009
9. 9
   Learning Depth-Aware Deep Representations for Robotic Perception

   31 citations · 2016
10. 10
    PoseFix: Correcting 3D Human Poses with Natural Language

    29 citations · 2023

Key Collaborators

Contact & Links