Humanoid

This category contains 169 ranked humanoid robots from 80 companies in 20 countries. The HMX is a specification score; legacy admission does not establish field-level review or independent task testing.

77 of the 169 records are at commercial or limited-production stage. These recorded maturity claims do not establish paid deployments or operating performance.

• Autonomy (0.25) — the single biggest open problem and value driver. A teleoperated robot is a puppet; a task-autonomous one is a worker. We deliberately make autonomy the heaviest pillar and penalise teleop-only systems hard, regardless of how slick the hardware looks in a demo.
• Manipulation (0.20) — useful work in human environments is bottlenecked on hands. Dexterous, high-DoF hands with real payload separate "walks around" from "does the job".
• Mobility (0.15) — table stakes. Bipedal balance is solved enough that it no longer dominates; speed/endurance still matter for duty cycles.
• Hardware (0.15) — total DoF, actuation modernity (electric > hydraulic for serviceability/quietness), payload-to-weight efficiency, onboard compute.
• Commercial (0.15) — does it actually ship and earn revenue? Real paid pilots (BMW, GXO, Mercedes, Amazon …) beat YouTube demos.
• Ecosystem (0.10) — capital, valuation and corporate backing decide who survives to iterate. Weighted lowest because money ≠ capability, but it is a real survival signal.