Help me choose Compare

How we collect, verify and score data

How HumanRobots.si collects and checks humanoid robot data: sources, provenance labels, regional landed prices, popularity and the score formula.

Where the data comes from

Every value links to its source: the maker's spec page or datasheet first, then launch posts, authorised resellers and press reports. Each robot page shows the date we last checked it. One record is one version that is sold separately, so "G1" and "G1 EDU" never share numbers.

Labels on values

LabelMeaning
OfficialMaker spec page or datasheet.
Official, approximateThe maker gives an approximate figure.
Maker claimFrom a launch event or post, not a spec sheet.
ReportedFrom press or analysts, not confirmed by the maker.
ResellerFrom an authorised reseller listing.
Our estimateEditorial estimate from comparable robots and reports.
CalculatedDerived from published figures, e.g. Wh = Ah × V.
Not disclosedThe maker does not publish it. We do not guess.

Estimates, reported and calculated values are never counted as wins in comparisons. Definitions that differ between makers are normalised: degrees of freedom include the hands, runtime is per charge (never "hours per day"), payload is the sustained two-arm carry, and one-off lift claims are listed separately.

Lifecycle status

  • Concept: Announced, nothing built yet.
  • Prototype: Working units, lab only.
  • Pilot deployments: Working shifts at customers.
  • Pre-order: Deposit now, ships later.
  • Shipping: Order and receive today.
  • Discontinued: No longer sold.

How we score

The score runs from 5 to 10. It is computed from the data, the same way for every robot, and recalculated when the data changes. Concepts and prototypes are not scored.

PartWeightWhat counts
Capability40%Joints, hands, payload, runtime, speed and autonomy.
Availability25%Lifecycle status and how many regions can order it.
Value20%Capability per dollar. Robots without a public price get a neutral value.
Support15%SDK openness, warranty, battery logistics and updates.

Scores by use

Every scored robot also gets a score for each use, on the same 5 to 10 scale. They weigh the same data for what each buyer needs, and the rankings for the home, research and industry use them.

UseWhat counts
HomeSafety and weight 22%, price 20%, availability 20%, autonomy 16%, hands 14%, voice 8%.
ResearchOpen SDK 30%, availability 20%, price 16%, joints 14%, hands 12%, simulation models 8%.
IndustrialPayload 22%, availability 20%, runtime and battery swap 18%, autonomy 18%, ratings and certifications 12%, units delivered 10%.
SocialVoice 26%, availability 20%, safety 18%, autonomy 16%, price 10%, hands 10%.

Popularity

The "#" in our tables is the popularity rank: robots are ordered by how often people search for them worldwide each month, measured with a keyword research tool and refreshed weekly. It says how much interest a robot gets, not how good it is; that is what the score is for. Robots without search data yet follow after the measured ones, in our editors' order.

Regional prices and availability

For each of six regions we record whether a robot ships locally, only by import, on pre-order, only to businesses, or not at all, with the seller and delivery time. A seller price includes VAT. Without one we estimate the landed price:

(US price + freight) × (1 + duty) × (1 + VAT) × exchange rate

  • United States: 1 USD = 1.00 USD, VAT 0.0%, duty 0.0%, freight $0
  • European Union: 1 USD = 0.86 EUR, VAT 21.0%, duty 2.7%, freight $900
  • United Kingdom: 1 USD = 0.75 GBP, VAT 20.0%, duty 2.0%, freight $900
  • China: 1 USD = 7.10 CNY, VAT 13.0%, duty 0.0%, freight $0
  • Japan and Korea: 1 USD = 1.00 USD, VAT 10.0%, duty 0.0%, freight $900
  • the rest of the world: 1 USD = 1.00 USD, VAT 15.0%, duty 5.0%, freight $1,200

Independence

No maker pays to be listed, scored or ranked. Some seller links earn a commission; they are marked and never change a score or a ranking. Makers can send corrections like anyone else, with a source.