What it is for
- Choosing a robot to build. Compare cost, parts, licenses and simulation models side by side, by robot type or in collections such as robots under $300.
- Research. Find robots that publish a URDF or MJCF model, or that link to LeRobot, and read the record as JSON through the API.
- Maintainers. Every page credits the project and links back to its source. Maintainers can claim their page to review and correct it.
Motari indexes the physical robot. Code stays on GitHub, and models and datasets on Hugging Face; Motari links out to them rather than copying them.
Where the data comes from
Every entry starts from the project's public sources: its repository, pinned to a commit, and the README, bill-of-materials files, documentation and build guides it publishes.
- Motari's ingest engine reads the repository and finds CAD files, URDF and MJCF models, meshes, bill-of-materials tables, firmware and license files.
- The 3D model on each page is rendered from the project's own URDF, CAD or mesh files and compressed for the web. The page says whether it covers the full robot, the main assembly or a simplified subset, and says so when there is no model yet.
- The catalog record adds the facts a builder needs: hardware, software and documentation licenses with the file that states each; the project's lifecycle (active, dormant, archived, deprecated, defunct); each published cost with its basis, scope and the date it was read; degrees of freedom; and the simulation models, pinned to the commit they were found at.
- Stars, forks, the last push and archived state come from the GitHub API.
How it is checked
- Every figure links to its source. What is not known is shown as unknown, never estimated.
- A cost is what the source publishes (a source total, an estimate or a vendor kit price), never a checkout Motari has verified, and it stays in the source's currency.
- Build steps are marked as following an upstream guide or as Motari's summary of a source, with the link either way.
- Pages are assembled by software agents from public sources, and each robot page says so, with the date it was assembled and when it was last re-verified against its sources.
- Reports that someone built a robot count only after a person reviews the evidence: the maintainer when the robot is claimed, otherwise a site admin.
- The reproducibility score comes from a published rubric; the methodology page explains each criterion and its weight.
Corrections, claims and removal
If you maintain a project in the atlas, you can claim its page with GitHub to review and correct it. Anyone can ask for a correction or for a page to be removed through the request form; removal requests are handled before other catalog work. Motari honors robots.txt.
Who maintains Motari
Details of the people and the legal entity behind Motari will be published on this page. Until then, the request form reaches the maintainers.
Source code
The ingest engine, command-line tool, SDKs and CI Action live in the Motari source repository, which is not public yet. The public JSON API is documented on the developers page.
Last updated 2026-09-30.