Google DeepMind Study: LLMs Control Robots in Physical World Without Fine-Tuning

m_wulfmeier · x · 2026-08-14

mwulfmeier shares work with Patrick Tresset exploring how LLMs (e.g., Gemini) without robot fine-tuning can control robots in the physical world. Published in IEEE RAS RAM, started at Google DeepMind, with learnings for prototyping at nomagicAI.

Related event: DeepMind: LLMs Control Physical Robots Without Fine-Tuning(2 posts)→

Original post →

More from Embodied

Embodied channel →