BotsGenius
ENNL
← Back to all companies
United KingdomAI & softwareCommercial

Google DeepMind

Bringing web-scale AI reasoning into the robot body.

Google DeepMind is Alphabet's consolidated AI research lab, headquartered in London. Its robotics work centres on Gemini Robotics, a family of vision-language-action models that extend the Gemini multimodal AI into physical manipulation, building on years of research through RT-1, RT-2, and the Open X-Embodiment collaboration.

Google DeepMind's robotics lineage runs from RT-1 and RT-2 — early demonstrations that large language and vision models could generalise robot policies — through the Open X-Embodiment dataset, which pooled manipulation demonstrations from dozens of research labs worldwide. Gemini Robotics applies the full power of the Gemini vision-language model to real-time robot control, enabling dexterous manipulation tasks that earlier systems could not handle. DeepMind has partnered with humanoid maker Apptronik to validate Gemini Robotics on bipedal hardware. The lab's research output also feeds the broader ecosystem through open datasets and published model weights.