AIToday
RoboticsWIRED AIPublished: Nov 8, 2026, 22:00 JST

Niantic Spatial uses Pokémon Go scans for robot training

Send on LINE
Niantic Spatial uses Pokémon Go scans for robot training

3 Key Points

  1. What happened

    Niantic Spatial CTO Brian McClendon says the company is using its trove of world data to reconstruct the real world in 3D, giving robots simulated training grounds where they can run millions of simulations.

  2. Why it matters

    By letting robots practice in realistic digital replicas instead of physical spaces, this approach could reduce the need for real-world testing and speed up how quickly robots adapt to new environments.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

Niantic's path to world models runs through Pokémon Go. The game, built around mixed-reality characters, generated scans of players' environments that the company now sees as a foundational data trove. After selling the game last year, Niantic pivoted to Niantic Spatial, with CTO Brian McClendon describing the mission as "very much about reconstructing the real world."

The technology is drawing interest beyond robotics. Wayve's Jamie Shotton says a world model must understand physics, cause and effect, and how other objects and people react to actions. Wayve uses dashcam footage to recreate hazardous roads that would be impractical to test in real life. Tencent is investing in world models for video games, where physics and 3D rendering are core.

Investors are placing large bets. Yann LeCun, Meta's former chief AI scientist, raised over a billion dollars for his Advanced Machines Intelligence startup, and AMD acquired Fei-Fei Li's World Labs for over $8 billion this year. Meanwhile, the data behind these models is being collected by gig workers, a labor market that may expand as companies seek more real-world video.

FAQ
What are world models?
Unlike text-based large language models, world models use visual inputs to generate digital environments that simulate physics, according to Oliver Cameron of Odyssey.
How do world models help robots?
They act as digital training grounds where robots can practice without physical movement, as Brian McClendon of Niantic Spatial describes, running millions of simulations to learn a space's parameters.
Who is investing in world models?
Yann LeCun raised over a billion dollars for his Advanced Machines Intelligence startup, and AMD acquired Fei-Fei Li's World Labs for over $8 billion this year.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleMicrosoft, CrowdStrike software demand grows: $5,000 split