Discussion Paper Version 0.2 rev9 · May 2026

HU-RSC – A classification system for the physical safety of smart robotic systems

Companion paper to HU-RAC under the shared HU-Codex umbrella. Addressed to the Humanide GmbH internally, to its service providers, suppliers and manufacturers, and to procurement functions at buyer organisations.

Author
Humanide GmbH, Cologne
Status
Public discussion paper — not a standard
Scope
Physical safety of smart robotic systems
As of
16 May 2026

Two models, one system

The Humanide classification system consists of two nominally independent, structurally parallel models with one asymmetric coupling: HU-RAC assesses digital sovereignty and provider resilience, HU-RSC assesses physical safety with respect to people.

Both models share the same architecture: a tier scale of 1 to 5, six axes, an aggregation principle and the expectations-matrix structure. The structural parallel is methodological, not aesthetic: buyers should be able to apply a single mental model. External communication uses a uniform phrasing — “this product reaches HU-RAC X / HU-RSC Y.”

Three assessment dimensions sit explicitly outside HU-RSC: digital sovereignty (the subject of HU-RAC), cybersecurity (HU-RAC matrix 5.3), and functional capability (Requirements Catalogue Part 5). HU-RSC focuses on personal safety in the sense of the ISO-12100 logic.

Seven use-case fields in scope

HU-RSC is calibrated to the robotics application fields that are commercially relevant in the DACH small- and medium-sized enterprise (SME) segment in the near term:

  • Consumer-facing robotics — lawn-mowers, vacuum and cleaning robots (B2C / SoHo)
  • Service robotics in semi-public spaces — hotels, hospitals, care facilities, schools
  • Industrial robotics in the SME segment — outside the classical safety-caged large-industry context
  • Outdoor robotics — agricultural, municipal, commercial outdoor areas
  • Humanoid robotics (pilot phase) — first SME pilot deployments
  • Assistance robotics — personal assistance, care support, edutainment
  • Mobile logistics robotics — AGV / AMR in SME logistics contexts

Specialist domains such as medical devices (MDR), nuclear installations or military robotics are outside the scope.

Where HU-RSC sits in the European framework

HU-RSC understands itself as a methodological orientation raster over several parallel pieces of legislation and standards — not as a competitor to any of them.

  • Machinery Regulation (EU) 2023/1230 — applicable from 20 January 2027. It carries the core obligations for persons placing machinery on the market and for operators with respect to physical safety.
  • AI Act (EU) 2024/1689 — high-risk obligations (risk management, data quality, transparency, oversight) apply to AI safety components.
  • Product safety and product liability — GPSR together with the revised Product Liability Directive 2024/2853.
  • Cyber Resilience Act (EU) 2024/2847 — update mechanics and minimum support period (Art. 13).
  • Safety standards — ISO 12100, ISO 13849-1, IEC 61508, ISO/TS 15066, ISO 10218-1/2:2025; in preparation: ISO 25785-1, ISO/IEC TS 22440.

HU-RSC is explicitly designed to be superseded by harmonised standards as they become available.

Six axes, five tiers

HU-RSC carries the same model architecture as HU-RAC: six assessment axes, a tier scale of 1 to 5, and the weakest-link principle as the aggregation rule. The tier scale measures the residual risk that remains after protective measures, not the original inherent risk.

The five tiers

Tier 1
Minimal residual risk
Stationary or very light mobile systems in a fully controlled environment; kinetically near-harmless; no safety-critical AI functions.
Desk-top robot arm for edutainment, simple vacuum robots without rotating tools, stationary inspection arms in a controlled industrial cell.
Tier 2
Limited residual risk
Light mobile systems in a semi-controlled environment; reversible harm is plausible, irreversible harm is unlikely. E_kin 1–10 J, no cutting or rotating tools.
Vacuum robot in a private home, simple AGVs with safety distances, inspection robots in commercial premises.
Tier 3
Elevated residual risk
Mobile systems with material kinetic energy or with cutting or rotating tools; human contact possible in a semi-controlled environment. E_kin 10–80 J.
Lawn-mowers with rotating blades in a private garden, cleaning robots in hotel corridors outside guest hours, AGVs in mixed human-robot environments.
Tier 4
High residual risk
Systems with material injury potential in a semi-controlled or public environment; vulnerable groups plausibly exposed; AI-driven decisions with safety relevance.
Humanoid pilot deployments in SME industry, cleaning robots in hospitals outside sensitive zones, outdoor lawn-mowers in publicly accessible parks.
Tier 5
Critical residual risk
Compound hazards with cumulative harm types, fatality potential, or direct and prolonged exposure of vulnerable groups without human ultimate control. Tier 5 is a qualitative step beyond Tier 4, not a quantitative one.
Care-assistance robots with physical contact to vulnerable persons; humanoid systems in unstructured environments with free third-party access.

The six assessment axes

  • 1

    Deployment environment

    Controlled (safety cage), semi-controlled (private home, hospital corridor, workshop) or public / unprotected (park, footpath). Secondary factors: light, floor conditions, weather, noise, visibility.

  • 2

    Form factor, motion envelope and kinetic injury potential

    Kinetic energy after E_kin = ½ m v² (orientation values from ISO/TS 15066), motion envelope, tools and contact surfaces, stability under emergency stop. For humanoids: the inverse-pendulum problem — stopping is not the same as being safe.

  • 3

    Harm types and harm severity

    Mechanical, thermal, chemical, electrical, biological, psychological — following the classification in ISO 12100 Table B.3. Secondary dimension: reversibility (fully reversible, long-term, irreversible).

  • 4

    Degree of autonomy

    Five-step: A0 (teleoperated), A1 (assisted), A2 (partly autonomous), A3 (highly autonomous with oversight), A4 (fully autonomous without human ultimate control). It measures the demand on the safety architecture, not its fulfilment.

  • 5

    Safety architecture — three pillars

    Pillar 1: classical functional safety (ISO 13849, IEC 61508). Pillar 2: AI safety architecture (AI-Act conformity, explainability, behavioural safety). Pillar 3: lifecycle update infrastructure (patchability, OTA, CRA conformity).

  • 6

    Exposed persons and exposure frequency

    Who is exposed, and how often? Only trained staff, adult private persons, mixed audiences, vulnerable groups (children, older people, patients). Brief encounter versus prolonged co-presence.

Aggregation logic

Overall placement follows the weakest-link principle: the strongest-loaded axis sets the overall tier. A single axis is enough — a humanoid robot with Tier-5 stand-and-fall potential on Axis 2 reaches Tier 5 overall, even if Axis 4 only reaches A2.

Within the three-pillar Axis 5, a modified minimum-floor rule applies: Pillar 3 (lifecycle update infrastructure) only enters axis aggregation from Tier 3 upwards, because CRA reach and the weight of the update obligation are marginal at Tier 1 and Tier 2. Pillars 1 and 2 are aggregated across all tiers under the weakest-link principle.

E-Min, E-Plus, R — graduated expectations per axis and tier

For each axis and tier, HU-RSC defines three expectation columns: E-Min (minimum essential, usually legally or market-standard expected), E-Plus (aspirational robustness uplift), R (residual-risk note as a duty-of-care warning). Six matrices — one per axis — together form the full profile. The matrix for Axis 6 (Exposed persons) shown as an example:

TierE-Min — Minimum measuresE-Plus — Robustness upliftR — Residual risks & warning
Tier 1Only trained staff exposed; brief encounter, no prolonged co-presence.Documented training history; attendance check.Check for reclassification if the audience changes.
Tier 2Adults exposed; no vulnerable groups; brief to moderate co-presence.Conduct rules documented; access by children and animals excluded.Pet encounters in a private setting cannot be ruled out.
Tier 3Mixed audiences possible; children and older people excluded or organisationally protected; moderate co-presence, plannable and bounded.Exclusion times for sensitive groups; audible or visual approach warning.Unannounced third-party presence cannot be fully ruled out — organisational duty of supervision.
Tier 4Vulnerable groups plausibly or regularly exposed; prolonged or frequent co-presence; documented supervision or protection plan.Adaptive behaviour by audience type; redundant person detection.Spontaneous reactions of vulnerable persons cannot be fully prevented; psychological consequences are a co-existing residual risk.
Tier 5Vulnerable groups directly and continuously exposed (care, child supervision); comprehensive protection plan with supervising staff; emergency presence guaranteed.Person-specific adaptation (e.g. patient profiles); continuous attendance monitoring.Continuous direct contact with vulnerable groups is regulatorily and ethically particularly sensitive — deployment only after explicit prior assessment by qualified experts.

The full set of six matrices — one per axis — is set out in the discussion paper. In HU-RSC, the R column is to be read as a duty-of-care warning, not as a disclaimer of liability.

Business criticality and provider resilience

The tier scale of 1 to 5 is the primary assessment dimension in HU-RSC. Two secondary dimensions complement it — parallel in structure to HU-RAC, reinterpreted in safety terms.

Level A · B · C

Business criticality

A — critical: a safety incident threatens the operator’s existence (operational shutdown, serious personal injury, market exit).

B — important: operationally important but bridgeable; harm is not existence-threatening.

C — tolerable: outage within tolerance, no material third-party impact.

R1 · R2 · R3

Provider resilience

R1: established corporate with long supply history.

R2: stable mid-cap with demonstrable continuity.

R3: start-up with elevated insolvency or change-of-control risk. At HU-RSC Tier 4 and above, defensible only with structured continuity mechanisms in place.

Reclassification upon substantial modification

HU-RSC is dynamic: a substantial modification within the meaning of Machinery Regulation (EU) 2023/1230 triggers a fresh tier assessment. Typical triggers are changes to tools, reach, degree of autonomy, AI components or the deployment context (extension to vulnerable groups, move from a controlled to a semi-public environment).

Reclassification is not an optional extension but a documented obligation in the procurement file — it ensures that the originally accepted tier still holds.

Minimum coupling HU-RSC × HU-RAC

There is an asymmetric minimum coupling between HU-RSC and HU-RAC: HU-RSC ≥ 4 requires HU-RAC ≥ 3.

Rationale: from Tier 4 (high residual risk) onwards, safety telemetry becomes safety-relevant — if logging, AI behavioural recording and update delivery are not integrity-protected, the measures expected under HU-RSC Pillars 2 and 3 do not carry the load assigned to them. A minimum HU-RAC capacity is therefore a precondition for HU-RSC Tier-4 / Tier-5 expectations to hold technically.

The coupling is asymmetric: a high HU-RAC does not require a high HU-RSC (a data-sovereign system need not be physically dangerous), but a high HU-RSC requires a sufficient HU-RAC.

Application in procurement practice

HU-RSC is designed primarily for procurement practice in the DACH SME segment. Three addressee groups are in view:

  • The Humanide GmbH internally — as the entity placing products on the market under its brands (Mowio, Wuffi, Robofixx, Robokitchen, Omtanke).
  • Service providers, distributors and manufacturers — who supply or co-deploy components and platforms.
  • Buyers and procurement functions at customer organisations, who evaluate Humanide products against their own internal requirements.

Application in four steps: (1) Use-case placement in one of the seven fields; (2) Tier determination over the six axes with the aggregation rule; (3) Derive requirements from the expectations matrices (E-Min as the minimum set, E-Plus as a discussion input); (4) Secondary dimensions A / B / C and R1 / R2 / R3 as context filters.

The tier classification must not be represented to suppliers as a certification — it is a procurement heuristic and does not replace the ISO-12100 risk assessment.

Discussion invited

HU-RSC is published as a discussion paper — not as a finalised model. Three model-building decisions in this version are explicitly marked as provisional: the selection of the six assessment axes, the choice of an HU-RSC-native tier scale with ISO mapping tables rather than direct aggregation of existing ISO classes, and the asymmetric minimum coupling HU-RSC ≥ 4 ⇒ HU-RAC ≥ 3. All three may be revised in v0.3 based on practitioner feedback.

Adopters should treat the statements in this paper as tendencies of the current model, not as final commitments. Feedback from procurement, engineering and standardisation work is expressly welcome.

Author and feedback

Humanide GmbH
Dittmar Müller, Managing Director
Mauritiussteinweg 11, 50676 Cologne, Germany

Email: info@humanide.com
Phone: +49 221 922840

Imprint

Information pursuant to Section 5 of the German Telemedia Act (TMG).

Provider

Humanide GmbH

Mauritiussteinweg 11
50676 Cologne
Germany

Warehouse and delivery address

Mauritiussteinweg 11
50676 Cologne
Germany

Represented by

Dittmar Müller
Managing Director

Contact

Phone: +49 221 922840
Email: info@humanide.com

Commercial Register

Local Court of Cologne (Amtsgericht Köln)
Registration number: HRB 115169
Registered office: Cologne

VAT Identification Number

VAT ID under § 27a UStG:
DE362951755

Privacy Notice

Information pursuant to Art. 13 GDPR on the processing of personal data on this website. Last updated: 30 April 2026.

1. Controller

The controller responsible for the processing of personal data on this website within the meaning of Art. 4 (7) GDPR is:

Humanide GmbH
Mauritiussteinweg 11, 50676 Cologne, Germany
Phone: +49 221 922840
Email: info@humanide.com

2. Data Protection Officer

Humanide GmbH has not appointed a Data Protection Officer, as the conditions of Art. 37 GDPR and Section 38 BDSG (German Federal Data Protection Act) are not met. For data protection inquiries, please contact the controller using the details above.

3. Nature of this Website

This website is a static technical publication. It does not use cookies, tracking technologies, web analytics, social media plugins, embedded videos, externally loaded fonts, or any other third-party services. No personal data is collected through forms, registration or comments — none of these features exist on this site.

4. Server Log Files

When you access this website, our hosting provider (IONOS SE, Elgendorfer Str. 57, 56410 Montabaur, Germany — acting as processor under Art. 28 GDPR pursuant to a data processing agreement) automatically records the following information in server log files:

  • IP address of the requesting device
  • Date and time of the access
  • URL of the accessed resource
  • HTTP status code and transferred data volume
  • User agent (browser and operating system identifier)
  • Referring URL (if transmitted by the browser)

These data are stored for a maximum of 8 weeks and are processed exclusively for ensuring the technical security and stability of the website (defence against attacks, troubleshooting). The legal basis is Art. 6 (1) lit. f GDPR (legitimate interest in operating a secure and stable website). The data are not merged with other data sources and are not used for the creation of usage profiles.

5. SSL/TLS Encryption

This website uses SSL/TLS encryption for the transmission of all content. You can recognise an encrypted connection by the "https://" prefix and the lock symbol in your browser.

6. External Links

This website contains links to external resources (e.g. www.humanide.com, hurac.eu, hucodex.eu, ec.europa.eu). When you click such a link, you leave this website and enter the data protection regime of the respective external operator. We have no influence on, and accept no responsibility for, the data processing on these external pages.

7. Your Rights as a Data Subject

Under the GDPR, you have the following rights regarding personal data concerning you:

  • Right of access (Art. 15 GDPR)
  • Right to rectification (Art. 16 GDPR)
  • Right to erasure (Art. 17 GDPR)
  • Right to restriction of processing (Art. 18 GDPR)
  • Right to data portability (Art. 20 GDPR)
  • Right to object (Art. 21 GDPR)
  • Right to lodge a complaint with a supervisory authority (Art. 77 GDPR)

The competent supervisory authority for Humanide GmbH is the Landesbeauftragte für Datenschutz und Informationsfreiheit Nordrhein-Westfalen (LDI NRW), Kavalleriestr. 2–4, 40213 Düsseldorf, Germany. To exercise your rights, please contact the controller using the details above.

8. Changes to this Privacy Notice

We may update this Privacy Notice to reflect changes in our processing activities or legal requirements. The current version, identified by the "last updated" date above, applies to all visits to this website.

Frequently asked questions

What is HU-RSC?

HU-RSC (Humanide Robotics Safety Class) is the safety pillar of HU-Codex. It evaluates the physical harm potential of smart robotic systems with respect to people across six axes, a five-step tier scale and the principle of the strongest burden. Status: Discussion Paper v0.2.11 (DE) and v0.2 rev9 (EN), May 2026 — explicitly not finalised.

Which axes does HU-RSC have?

The provisionally selected six axes are: type and mobility, kinetic energy, autonomy and decision certainty, human–robot interaction, functional safety of AI components and lifecycle safety. This selection is explicitly marked as provisional and may be revised in v0.3 based on practitioner feedback.

What does the coupling rule "HU-RSC ≥ 4 requires HU-RAC ≥ 3" mean?

The two pillars are nominally independent but asymmetrically coupled: if a robotic system reaches HU-RSC 4 or 5 — i.e. elevated physical residual risk —, HU-Codex requires an HU-RAC rating of at least 3 for the same system. The reverse does not hold. The coupling is defined as an asymmetric minimum relation in the HU-RSC Discussion Paper v0.2.

Does HU-RSC replace a risk assessment to ISO 12100?

No. HU-RSC is an aggregation heuristic for procurement and supplier dialogue, not a safety-engineering tool. An HU-RSC tier classification does not replace a risk assessment to ISO 12100, a functional safety assessment to ISO 13849 or IEC 61508, or a conformity assessment under the Machinery Regulation. It is a preliminary or accompanying layer, not a substitute.

Can a small robot still reach a high HU-RSC tier?

Yes. Size or weight alone do not decide. What matters is the most critical burden dimension — for example degree of autonomy, deployment environment, human–robot interaction or kinetic harm potential relative to exposed persons. A small robot in a busy care environment can reach a higher HU-RSC tier than a larger robot in a fenced industrial setting.

Which use-case fields are in the focus of HU-RSC?

Seven application fields with near-term commercial relevance in the DACH SME segment: consumer-grade robotics (lawn-mower, vacuum, cleaning robots), service robotics in semi-public spaces (hotels, hospitals, care, schools), industrial robotics in SMEs outside classical safety cages, outdoor robotics (agriculture, municipal, commercial outdoors), humanoid robotics (pilot phase), assistive robotics (care, edutainment) and mobile logistics robotics (AGV/AMR in SMEs). Specialist domains such as MDR medical devices, nuclear facilities or military robotics are explicitly out of scope.

How does HU-RSC relate to the Machinery Regulation and the AI Act?

HU-RSC uses the Machinery Regulation (EU) 2023/1230 (binding from 20 January 2027) as its central reference for market placement and operator obligations regarding physical safety. For AI safety components, the high-risk obligations of the AI Act (EU) 2024/1689 also apply. Further methodological frameworks include the Product Liability Directive 2024/2853, the GPSR, the Cyber Resilience Act 2024/2847 (update mechanism and minimum support period under Art. 13), and the safety standards ISO 12100, ISO 13849-1, IEC 61508, ISO/TS 15066 and ISO 10218-1/2:2025.

Why is HU-RSC called a "discussion paper" and not a "standard"?

HU-RSC is explicitly published as a discussion paper because three model-building decisions are not yet final: the selection of the six assessment axes, the choice of an HU-RSC-native tier scale with ISO mapping tables rather than direct aggregation of existing ISO classes, and the asymmetric minimum coupling HU-RSC ≥ 4 requires HU-RAC ≥ 3. All three may be revised in v0.3 based on practitioner feedback. HU-RSC is also designed to be aligned with harmonised standards and, where appropriate, to be replaced or supplemented by them once these become available.

What is explicitly outside HU-RSC?

Three dimensions are explicitly outside HU-RSC: digital sovereignty (covered by HU-RAC), cybersecurity (HU-RAC matrix 5.3) and functional capability (Part 5 of the requirements catalogue). HU-RSC focuses on the protection of persons in the sense of the ISO 12100 logic. Also outside the scope: MDR medical devices, nuclear facilities, military robotics.