Methodology
Core verification workflow:
Every claim is traceable to an exact source quote and preserved artifact. Conflicting evidence is published, not hidden. Every published profile has passed human review.
What the badges mean
Every badge is a deterministic, explainable label computed from the evidence actually linked to a claim — never a hidden or weighted score.
Verification status
In one line: we mark a claim "Verified" when at least one independent, reputable source explicitly supports it and no equally strong source directly contradicts it.
- Verified
- At least one Manufacturer-tier source and at least one Independent-tier source both support the claim — both sides agree.
- Non-manufacturer finding
- Independent or Supporting-tier evidence exists, but the manufacturer hasn't stated this itself (e.g. a third-party report on a limitation the manufacturer never addressed). Deliberately not called "Independent finding" — this bucket also covers Supporting-tier-only evidence (overview/database aggregator sites), which isn't held to the same rigor as a true Independent source (press, analyst, researcher), so the label doesn't overclaim.
- Unverified claim
- Only the manufacturer has said this — no independent source has confirmed or denied it yet.
- Contradicted
- A genuine value disagreement exists between sources (not a caveat or an ambiguous read) — see the open or resolved mismatch on that claim for the specific conflicting quotes.
Confidence
- High
- 2 or more Manufacturer/Independent-tier sources support the claim, including at least one Independent source. Supporting-tier (overview/database) sources are deliberately not counted toward High — those sites often just republish the manufacturer's own figures.
- Medium
- At least one source supports the claim, but not enough to reach High.
- Low
- No supporting evidence is currently linked.
- Disputed
- An open mismatch exists on this claim — a reviewer hasn't resolved a conflict in the evidence yet.
Robot classification
The "Robot classification" section on each profile follows the multi-axis category model from ISO 8373:2021 (general robot vocabulary), ISO 10218 (industrial robots and applications), ISO 13482 (personal-care robots), and the draft ISO/CD 25785-1 (dynamically stable industrial mobile robots). A real robot is typically several of these categories at once — a legged warehouse robot, for example, can be a mobile robot, an industrial robot, and a dynamically-stable robot simultaneously — so each axis (mobility, manipulator type, environment, stability, and so on) is assessed independently rather than collapsing the robot into one "type" label. Where a source doesn't address a given axis, it's shown as "Undisclosed" rather than guessed.