Introducing Rayform

Today we are opening the waitlist for Rayform, a licensed dataset of the physical world built for the teams training robots.

The gap

Language models had the whole internet to learn from. Image models had every photograph ever uploaded. Robot models have a handful of lab datasets: a few rooms, a few objects, usually one gripper, usually one camera.

That is not enough to learn a world. A robot that will work in a kitchen, a warehouse, or a mine has to understand space, motion, and consequence in places it has never seen. The data to teach it that barely exists, and what exists was mostly recorded for something else.

What Rayform is

Rayform records real environments the way robots need them: in three dimensions, over time, across the places where robots will actually work. Every capture carries the same five layers, aligned frame by frame:

• Scenes. Dense 3D reconstructions of the space, with the raw depth and LiDAR frames behind them.
• Motion. Camera, object, and body trajectories through that space, with a full pose on every frame.
• Interaction. People and robots acting in the scene: grasps, placements, doors, drawers, tools.
• Robot state. Joint positions, velocities, and gripper state, in sync with every sensor frame.
• Context. Where, when, and under what conditions it was captured.

The result is one corpus, not a pile of recordings. A kitchen captured in one city and a kitchen captured in another share a coordinate convention, a clock, and a vocabulary, so you can ask for every door opened in low light and get exactly that.

Who it is for

Any team training a model that has to act in the physical world: foundation-model labs, humanoid and manipulation companies, navigation and autonomy teams, drone operators, and research labs. The same captures serve a humanoid learning to grasp and a drone learning to survey a mine.

What the preview is

Rayform is in private preview. We are onboarding a small number of teams at a time, working closely with each one on the environments and tasks they need, the formats their pipeline expects, and the terms of their licence. What preview teams ask for shapes what we capture next.

If you are training robots, join the waitlist. Tell us what you are building, and we will show you what Rayform already holds for it.