معرفی
Boyuan Chen is a research scientist at OpenAI and MIT's Department of Electrical Engineering and Computer Science (EECS), specializing in world models, reinforcement learning, and embodied AI. His work bridges AI systems with physical world understanding through innovative approaches to video generation, spatial reasoning, and robotic control.
- PhD in EECS from MIT with a minor in philosophy
- Bachelor's degree in computer science and mathematics from UC Berkeley
His research focuses on:
- World Models: Developing architectures like Diffusion Forcing Transformer to enable stable long-horizon video generation and planning
- Embodied AI: Advancing spatial reasoning capabilities in Vision-Language Models (SpatialVLM) for robotics applications
- Reinforcement Learning: Creating self-supervised methods for cross-task transfer learning and soft robot control
Key article trends:
- 2025: History-Guided Video Diffusion for flexible temporal conditioning
- 2024: Diffusion Forcing combining next-token and sequence-level modeling
- 2024: SpatialVLM introducing 3D spatial reasoning to vision-language models
- 2023: Self-supervised RL transfer using random features
- 2021: 3D keypoint learning for robotic control
Advising & Collaborations:
- PhD advisors: Prof. Russ Tedrake and Prof. Vincent Sitzmann (MIT)
- Undergraduate mentor: Prof. Pieter Abbeel (UC Berkeley)
- Industry internship: Dr. Fei Xia (DeepMind)
Robotics & Team Leadership:
- Co-founder of MIT's Robomaster ICRA challenge team
- Founder of high school's first FRC team
- Developer of personal robots like Autonomous Bogie Rover and Autonomous Drone
۰مقاله منتشرشده
