01
Developer Experience
How quickly can a competent engineer make it do something new?
Public SDKAPI documentationWorked examplesROS 2 supportSimulation modelRelease frequency
A single number from 0 to 100 that summarises seven measured dimensions of a robot. It is arithmetic, not judgement: the dimensions are published, the weights are published, and the calculation on this page is the same code that produces the number on every robot page.
The platform never says one robot is better than another. It says one robot scores higher, and shows you why.
Example · A0509 — same arithmetic as the worked example below.
Each dimension answers one question about living with the robot, and each is derived from evidence we can point at rather than from the specification sheet alone.
How quickly can a competent engineer make it do something new?
Can it be deployed on a site rather than demonstrated in a lab?
Can a lab get beneath the vendor's abstraction layer?
What exists around the robot that you do not have to build?
What happens on the day a part fails?
Where can it physically go, and for how long?
How much can it carry and control that the vendor did not ship?
All weights, formulas and data sources are published. The same calculation is used across the platform, today and in the future.
A dimension that cannot apply is left out rather than scored zero. A bench-mounted arm has no Mobility, and pretending otherwise would punish it for being the kind of robot it is.
The published score is the unweighted mean of the dimensions that apply, rounded to the nearest whole number. Equal weighting is the only split that needs no defence; anything else would be a preference presented as a measurement.
| Dimension | Score | Weight | Product |
|---|---|---|---|
| Developer Experience | 92 | ×1 | 92 |
| Industrial Readiness | 51 | ×1 | 51 |
| Ecosystem | 91 | ×1 | 91 |
| Maintainability | 55 | ×1 | 55 |
| Expandability | 35 | ×1 | 35 |
| Total | 5 | 324 |
324 ÷ 5 = 64.80 → 65
A0509 has 2 dimensions we do not score, so the mean is taken over 5 rather than 7.
The same evidence answers different questions depending on who is asking. Change the weighting and the arithmetic above changes with it — which is the clearest demonstration that nothing here is a judgement call.
Every dimension counts the same. This is the number shown on robot pages.
The bands exist so a score can be described in a sentence. They are labels on a continuous scale, not categories.
Four things a robotics buyer might expect to find here, and why they are deliberately absent.
A score that moves when a comment section gets busy is a popularity measure wearing a lab coat. Reviews may appear on robot pages; they never enter the score.
Price is a fact we record and let you filter on, not a virtue. A cheaper robot is not a better one, and cost changes by region, quantity and year.
A figure in a launch video is not evidence. Specifications enter the catalogue from manufacturer documentation, cited on the page they appear.
No editor can override a score. If we think the result is wrong, the fix is to correct the evidence or the method in public, not to nudge the number.
Every specification carries one of three states, shown on the robot page next to the value itself.
Found in the manufacturer's own documentation and linked from the specification row.
The manufacturer has not published this figure. The row stays empty rather than being filled with an estimate.
The figure circulates but we could not trace it to a primary source, so it is marked and excluded from scoring.
The honest limits of a single number.
It cannot tell you whether a robot suits your task. A quadruped scoring 82 and a cobot scoring 82 are not interchangeable; the score compares robots against the same seven questions, not against your requirement.
It cannot see what has not been published. A manufacturer that documents nothing scores badly on dimensions it might deserve to win, and we would rather understate a robot than invent a figure for it.
It moves. When a vendor publishes an SDK or a certification, the evidence changes and so does the score. Every change is dated, and a robot page shows what changed and when.