- Gemini Robotics 1.5: Pushing the Frontier of Generalist Robots
- RT-Affordance: Affordances are Versatile Intermediate Representations for Robot Manipulation
- Vision Language Models are In-Context Value Learners
Awards: IEEE ICRA 2024 Best Conference Paper Award
Projects: Gemini Robotics series (e.g., Gemini 2.0, Gemini 2.5, etc.)
Research Experience
Position: Staff Research Scientist at Google DeepMind
Teams: Robotics team and Gemini multimodal team
Work Experience: Founded Machine Learning at Berkeley before joining Google DeepMind
Education
Degrees: B.S. and M.S. in Electrical Engineering and Computer Science
University: UC Berkeley
Advisor: Professor Claire Tomlin
Background
Research Interests: Robot learning, internet-scale foundation models, and reinforcement learning. Focuses on methods that can scale and generalize in the real world.