DeepMind trains Apollo humanoid from video using Gemini 4 Argon
Google DeepMind is using Gemini 4 Argon to teach Apptronik’s humanoid robot, Apollo, physical skills directly from video recordings.
What happened?
Google DeepMind has demonstrated how Gemini 4 Argon and its robotics variant can teach the Apptronik humanoid robot, Apollo, new physical skills directly from video. The model observes a task in video footage, extracts the motion patterns, and transfers them to the hardware without the need for separate teleoperation.
Key facts
| Modell | Gemini 4 Argon (robotikvariant) |
|---|---|
| Robotsystem | Apptronik Apollo |
| Träningsmetod | Inlärning via videoanalys |
Why it matters
Training via video addresses a major bottleneck in physical AI: the requirement for manually collected teleoperation data for every new task. If robots can learn directly from video recordings, the cost and time required to scale humanoid systems could decrease significantly.
Who is affected?
The technology is relevant for robotics developers, industrial companies, and automation engineers seeking to scale the training of general-purpose robots without expensive manual data collection.
What else you should know
DeepMind also demonstrated how the same model can create a 3D cathedral in Blender from a single image using the Model Context Protocol (MCP), illustrating the model's versatility in tool usage. This development aligns with initiatives such as Nvidia's Cosmos and Hugging Face's robotics projects.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Vilken robot användes i demonstrationen?
The link opens in a new window and leads to the publisher's own site.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "DeepMind trains Apollo humanoid from video using Gemini 4 Ar"