Papers
11
Total Citations
80
H-Index
6
About
Andrei Barbu is a leading researcher at the intersection of computer vision, robotics, and natural language understanding, with a core focus on building machines that can learn and reason like humans. His work is unified by the challenge of grounding language in physical and social reality, enabling robots to understand commands, learn from observation, and engage in complex interactions. A key contribution is his pioneering work on **physically-instantiated game play**, where a robotic system learns to play board games by watching human demonstrations (23 citations). He has also made significant strides in **systematic generalization for grounded language**, demonstrating how compositional neural networks can understand novel combinations of concepts (13 citations). Barbu’s research extends into **social reasoning**, formalizing social interactions as recursive Markov decision processes to give robots social skills (7 citations). His notable achievements include developing a natural-language parser that translates commands into linear temporal logic for grounded robotics, and creating a visual language model for estimating object pose and structure. With over 80 citations across his most-cited works, Barbu’s innovative approach—blending vision, language, and social cognition—is shaping the future of intelligent, socially-aware robots.
Research Focus
Key Achievements
Top Papers
- 1Learning physically-instantiated game play through visual observation23 citations · 2010
- 2
- 3Partially Occluded Hands:7 citations · 2019
- 4Social Interactions as Recursive MDPs7 citations · 2021
- 5Incorporating Rich Social Interactions Into MDPs6 citations · 2022
- 6
- 7
- 8
- 9Compositional RL Agents That Follow Language Commands in Temporal Logic3 citations · 2021
- 10Seeing Unseeability to See the Unseeable3 citations · 2012