Leonardo Barcellona
Papers
5
Total Citations
53
H-Index
3
About
Leonardo Barcellona is a robotics researcher advancing the frontier of human-robot collaboration and intelligent manipulation. His work centers on three key areas: action and gesture recognition for safe human-robot interaction, semantic grasping with minimal training data, and robotic waste sorting. Barcellona’s most impactful contribution is a general skeleton-based framework for recognizing human actions and gestures in collaborative settings (41 citations), enabling robots to anticipate human movements during tasks like assembly. He also developed FSG-Net, a deep learning model that achieves semantic robot grasping through few-shot learning, allowing robots to grasp specific objects with only a handful of examples. In waste management, Barcellona created WasteGAN, a data augmentation method using generative adversarial networks to improve robotic sorting of highly variable waste items. His recent work on compositional world models empowers robots to imagine and imitate realistic behaviors, reducing hallucinations in learned policies. With publications spanning 2021 to 2024, Barcellona’s research directly addresses the perception and manipulation challenges that limit real-world robot deployment, making him a rising voice in applied robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Multi-view Human Parsing for Human-Robot Collaboration3 citations · 2021
- 4
- 5