Ranked on what ships.
107 robots scored out of 100 on three questions: is it working for paying customers, is it doing so without a person on a controller, and can you buy one. Demos, renders and announcements score zero.
How the score is built.
Deployment (0 to 50): units at customer sites, number of named customers, duration. Autonomy (0 to 30): what has been shown running without teleoperation, and whether someone independent saw it. Availability (0 to 20): a public price, an order button, a delivery date that held.
Full rubricReading the table
Retired and research-only platforms are excluded; they appear in the robot database and the timeline. A robot enters the Index when its maker calls it a product, or when it is running at a customer site. Scores change when evidence changes, not when press releases do.
Where a maker publishes numbers we cannot verify, we use them and mark the estimate. Where nobody publishes numbers, the deployment score stays low. That is deliberate: the absence of evidence is information too.
Disagree with a score? Send evidence. Public, sourced evidence moves scores within a day.