Papers
9
Total Citations
519
H-Index
8
About
Javier Romero is a pioneering researcher at the intersection of computer vision, robotics, and human-robot interaction, with particular expertise in grasp recognition, hand pose estimation, and imitation learning. His most influential work, "Visual Object-Action Recognition: Inferring Object Affordances from Human Demonstration" (2010, 236 citations), established a compelling framework for robots to learn object manipulation by observing human actions—a foundational contribution to Programming by Demonstration systems. Building on this, Romero developed sophisticated methods for translating human grasps into executable robot commands, addressing the deceptively complex challenge of bridging anatomical differences between human and robotic hands. His 2012 metric for comparing anthropomorphic motion capability of artificial hands (92 citations) provided the field with a much-needed benchmarking tool for prosthetic and robotic hand design. He further advanced the representation of hand motion through nonlinear postural synergies and tackled marker-less robot pose estimation using depth-image classification—work that has influenced both industrial robotics and rehabilitation engineering. With over 500 cumulative citations, Romero's research has meaningfully shaped how robots perceive, learn from, and replicate human dexterous manipulation, making him a significant voice in embodied AI and robotic learning communities.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Robot arm pose estimation through pixel-wise part classification56 citations · 2014
- 4Extracting Postural Synergies for Robotic Grasping53 citations · 2013
- 5Visual recognition of grasps for human-to-robot mapping38 citations · 2008
- 6Grasp recognition and mapping on humanoid robots18 citations · 2009
- 7Human-to-Robot Mapping of Grasps12 citations · 2008
- 8Modeling and evaluation of human-to-robot mapping of grasps8 citations · 2009
- 9From Human to Robot Grasping6 citations · 2011