Industry Research

Appearance

EU ADS Type Approval: 2022/1426 and 2026/481 ​

Last updated: 2026-05-09

Why It Matters ​

The EU has one of the clearest upfront approval structures for fully automated road vehicles. Implementing Regulation (EU) 2022/1426 created uniform procedures and technical specifications for type-approval of automated driving systems (ADS) in fully automated vehicles. Implementing Regulation (EU) 2026/481 amends that framework and adds an automated valet parking (AVP) path that begins lifting the earlier small-series constraint for that use case.

For deployment teams, the EU model is valuable even outside Europe because it defines what a regulator expects in a serious ADS dossier: ODD, safety concept, scenario validation, minimal-risk behavior, cyber/software-update governance, data recording, in-use monitoring, and operational manuals.

Evidence/Map ​

Topic2022/1426 baseline2026/481 amendmentPractical implication
ScopeType-approval of fully automated M and N category vehicles with regard to ADS. Covered use cases include predefined-area transport, hub-to-hub routes, and automated valet parking.Keeps the 2022/1426 architecture but amends definitions and adds AVP-specific Annex V.Start by classifying the exact ADS feature and ODD. Approval is not generic "self-driving"; it is feature and ODD specific.
Safety targetADS must be free from unreasonable safety risks to occupants and other road users in the relevant ODD.AVP gets additional safety performance and validation requirements because the use case is low-speed parking-area operation.Safety claim needs a comparator, acceptance criteria, and validation targets, not just miles driven.
Scenario evidenceRequires nominal, critical, and failure scenario selection and validation evidence.AVP Annex V adds AVP-specific definitions such as AVP trip, transition location, smallest relevant object, safety distance, permanent/static/dynamic objects.Build scenario libraries around ODD boundaries and operational insufficiencies, including pedestrians, occlusion, infrastructure, and failures.
MRM/MRCDefines minimal risk manoeuvre and minimal risk condition.AVP requirements preserve the need for safe fallback behavior in the parking ODD.Every EU-style dossier needs a clear degraded-state and stop/recovery story.
Remote interventionRemote intervention operator may perform allowed support tasks but does not drive the vehicle; ADS continues to perform the dynamic driving task.AVP may use external infrastructure, but the ADS feature remains bounded by the ODD and technical specifications.Avoid describing remote assistance as remote driving unless the legal/technical model actually allows direct control.
Cyber and software updatesInformation document requires cyber-security type approval, cyber-security management system certificate, software update type approval, software-update management certificate, and ADS software identification.Safety management obligations were clarified to include arrangements with organizations involved in development, manufacturing, or in-use deployment.Supplier interfaces and software version traceability are approval artifacts, not back-office details.
Scale2022 framework operated with small-series limitations for fully automated vehicles.2026/481 recitals state the small-series limitation is lifted first for AVP; AVP feature active speed must not exceed 30 km/h.Do not read AVP scale-up as unlimited approval for all robotaxi or hub-to-hub use cases.

Practical Use ​

Build the EU ADS dossier around these work products:

  • ODD definition: geography, infrastructure, speed, weather, lighting, road/parking geometry, traffic participants, and external infrastructure dependencies.
  • ADS architecture: sensors, compute, actuators, maps, localization, backend, remote capabilities, and safety-relevant interfaces.
  • Safety concept: hazards, unreasonable-risk argument, malfunction behavior, operational insufficiency handling, ODD boundary handling, MRM/MRC triggers, and recovery.
  • Scenario plan: nominal, critical, and failure scenarios with traceability to ODD, OEDR, traffic rules, VRUs, infrastructure, and environmental conditions.
  • V&V evidence: simulation, track/lab tests, real-world tests, tool credibility, uncertainty, pass/fail criteria, and interpretation of results.
  • Cyber/software package: UN cyber/software approvals or equivalent certificates, software identification, update eligibility, rollback/version controls, and supplier controls.
  • Data and in-use monitoring: stored data elements, access, security, privacy, field performance, incident learning, and type-approval authority reporting.
  • Operating manual: roles of owner, transport service operator, onboard operator, remote intervention operator, maintenance, environmental restrictions, and failure instructions.
  • AVP-specific addendum: transition locations, parking ODD, smallest relevant object, safety distance, object classes, traffic flow impact, and <=30 km/h active AVP speed claim.

Failure Modes or Caveats ​

  • EU type approval is scoped to vehicle type, ADS feature, software identity, and ODD. Changes to software, sensors, ODD, remote operations, or external infrastructure can trigger reassessment.
  • Member States are not required to predefine areas, routes, or parking facilities under 2022/1426 and may still regulate circulation and local transport service operation.
  • The 2026/481 scale improvement is AVP-specific. It should not be used as evidence that all fully automated vehicle use cases are ready for unlimited series approval.
  • "Remote intervention" in 2022/1426 is not remote driving. The ADS remains responsible for the dynamic driving task.
  • EU AI Act obligations can apply in parallel where the autonomy system or safety components fall within high-risk AI/product safety rules.
  • Approval evidence must include both functional safety and operational safety; testing only nominal driving is not enough.

Sources ​

Public research notes collected from public sources.

Copyright © 2026 kvynlim