Faculty Profile: Prof. Yonatan Bisk — CMU LTI + Robotics
Faculty Profile: Prof. Yonatan Bisk — CMU LTI + Robotics
Position: Assistant Professor (LTI + Robotics Institute) Institution: Carnegie Mellon University Lab: REAL Center (Robotics, Embodied AI, and Learning) Website: https://www.lti.cs.cmu.edu/people/faculty/bisk-yonatan.html Report date: 2026-06-12
Research Focus
Embodied AI, multimodal language grounding, VLM for robotics/physical tasks, multi-party human-robot-language interaction, on-device ML.
Academic Profile
- Joined CMU ~2020
- Amazon Research Award (2025)
- DARPA ISAT Study Group member (2024)
- Co-founder of REAL Center at CMU
- Teaches: “Talking to Robots”, “Multimodal Machine Learning”, “On-Device Machine Learning”
Key Publications
- SOTOPIA (co-author, ICLR’24)
- Embodied AI + LLMs survey work
- Multimodal grounding and vision-language for physical tasks
Fit with Weijia Zhang
| Dimension | Assessment |
|---|---|
| VLM / multimodal | ✅ Core focus |
| Embodied agents | ✅ Primary research area |
| Language grounding | ✅ Strong |
| GUI agents (non-physical) | ⚠️ Physical robots primarily, not GUI |
| RL post-training | ⚠️ Robot learning uses RL but not LLM RL training |
| NLP agents | ⚠️ Adjacent |
Verdict
P2 套磁（如果你想往 VLM embodied agent 方向走）。 If Weijia’s GUIAgentDebugger evolves toward physical/embodied agents or multimodal grounding, strong match. Less relevant for pure NLP/GUI agent direction.