Methodology v0.1

Evidence has levels.

Useworthy is designed to show how strongly a robot claim is supported. The alpha does not issue certification and deliberately avoids a single overall score when long-term data is missing.

E1

Manufacturer claim

Datasheets, manuals, product pages and vendor statements.

E2

Public demonstration

Observable demo or company-reported field result without independent verification.

E3

Independent test

A repeatable third-party test with documented conditions and product version.

E4

Verified deployment

Operating data from a real deployment with evidence sufficient to verify the metric.

E5

Longitudinal fleet data

Multi-site, multi-unit, long-term field data with version and exposure normalization.

Metrics Useworthy is building toward

DimensionPriority metricsWhy it matters
ReliabilityHuman interventions / 100 productive hours; MTBF; MTTR; unplanned downtimeShows whether the robot removes work or creates supervision work.
AutonomyAutonomous task ratio; teleoperation ratio; recovery successSeparates a successful demo from truly autonomous work.
SafetyCollision force; fall behavior; e-stop latency; sensor/communication failure behavior; vulnerabilitiesCE status alone does not explain real-world application risk.
UsabilityTime-to-useful; setup hours; weekly maintenance; support responseIntegration burden can dominate purchase economics.
EconomicsCost per productive hourNormalizes purchase price, downtime, human assistance, maintenance and software.

Autonomy labels

  • A0Fully teleoperated
  • A1Continuous human supervision
  • A2Autonomous execution with frequent intervention
  • A3Single-task autonomous; human help mainly for exceptions
  • A4Multi-task continuous autonomy
  • A5Long-duration open-environment autonomy with rare intervention