Provider
Omnex
Duration
3 days
Access
Private / corporate · Runs on request
Delivery
Face-to-face
Sessions and enrolment
Check the provider's website for upcoming sessions and enrolment information.
Price
Price not published. Check with the provider.
Facilitator
Not identified in public sources checked
International delivery
Global delivery available

About this course

Covers the safety of vehicles whose driver assistance or automated driving systems contain AI sub-systems, using the requirements of ISO/PAS 8800. It starts with the specification's conformity requirements and tailoring, then applies ISO 26262 tailoring to machine-learning systems and examines how AI systems interact with the systems around them and where AI errors come from. Later chapters cover the AI safety lifecycle and its alignment with ISO 26262 and SOTIF, structured assurance arguments in GSN or CAE, AI-specific safety requirements, input spaces and safety KPIs, architectural patterns and measures, the dataset lifecycle, verification and validation and the sufficiency of AI testing, root-cause and fault analysis using FMEA or STPA, operational monitoring and re-approval, and confidence in AI toolchains. Each chapter has a breakout exercise, and the course ends with an exam.

Facilitators

Facilitator not identified in public sources checked.

Who should attend

  • Functional safety engineers, AI and machine learning engineers in automotive systems, systems engineers and safety architects, quality and compliance engineers, verification and validation specialists, and safety and project managers

Prerequisites

  • Familiarity with ISO 26262 and SOTIF, an understanding of AI and machine learning in safety-critical systems, and a background in vehicle development and risk analysis are helpful

Learning outcomes

  • Apply ISO/PAS 8800 to safety-related E/E systems that use AI technology.
  • Identify the risk of undesired safety-related behaviour at vehicle level.
  • Select AI architectures, technologies and measures for a typical project.
  • List and discuss the safety analysis elements of AI systems.
  • Understand AI safety requirements in detail.
  • Understand the principles behind a project-specific assurance argument for safety.

Course outline

  1. 01

    Day 1

    ISO/PAS 8800 conformity requirements and tailoring · AI system safety engineering · AI safety lifecycle · Assurance arguments (GSN or CAE)

  2. 02

    Day 2

    AI safety requirements, input space and KPIs · Architecture and measures · Dataset lifecycle · Verification and validation of AI

  3. 03

    Day 3

    Safety analysis of AI failures (FMEA, STPA) · Operational measures, monitoring and re-approval · Confidence in tools · Final exam

What participants receive

  • Course manual
  • Workbook with all breakout exercises

Training areas