Overview
ISO/TR 22100-5:2021 - Safety of machinery - Relationship with ISO 12100 - Part 5: Implications of artificial intelligence machine learning - is a technical report from ISO that guides how artificial intelligence (AI) and machine learning (ML) affect machinery safety. It explains how hazards associated with AI/ML applications in machinery can be considered within the ISO 12100 risk assessment and risk reduction process. The report applies where AI/ML is designed to act within specified limits (predetermined boundaries) and does not cover systems where AI acts beyond specified limits or safety-related parts of control systems.
Key topics and requirements
- Terminology and scope: Definitions of artificial intelligence and machine learning and clarification of where the report applies.
- AI lifecycle considerations: Recognition of the two stages-training and inference (autonomous operation)-and their implications for safety.
- Use limits and boundaries: Emphasis on defining intended use, space/time limits and other operational boundaries as part of the risk assessment.
- Risk assessment per ISO 12100: How to identify hazards, estimate and evaluate risks introduced or changed by AI/ML, and select risk reduction measures.
- Risk reduction strategies: Comparison of traditional guards versus sensor-based detection and control (e.g., adjust speed, stop, or reroute based on detection), and the need to adapt detection fields and controls to variable speeds/behaviour.
- Examples and case studies: Concrete examples showing when AI/ML does not add safety risk (e.g., precision spraying, automated part removal) and when it does (e.g., autonomous AGVs operating without perimeter safeguards).
- Residual risk and information for use: Requirement to evaluate residual risk after mitigation and provide appropriate user information.
Practical applications and users
Who should use this standard:
- Machine designers and OEMs integrating AI/ML into machines
- Functional safety and risk assessment engineers
- System integrators deploying AGVs, robotics, or smart manufacturing equipment
- Industry practitioners in agriculture (precision farming), manufacturing (laser cutting, part handling), logistics (autonomous guided vehicles), predictive maintenance and condition monitoring
- Standards committees and compliance/regulatory professionals
How it’s used:
- To guide risk assessments and safety-by-design when AI/ML alters machine behaviour within defined limits
- To decide when sensor-based protective measures are needed instead of fixed guards
- To document intended use, limitations and information for users to address residual risks
Related standards
- ISO 12100 - General principles for design, risk assessment and risk reduction
- ISO 3691-4 - Driverless industrial trucks (AGVs)
- ISO 17757 - Autonomous and semi-autonomous earth-moving/mining machinery
- ISO/IEC 2382, ISO/IEC 38505-1 - Relevant terminology and IT governance vocabularies
Keywords: ISO/TR 22100-5:2021, safety of machinery, AI, machine learning, risk assessment, ISO 12100, AGV, predictive maintenance, precision farming.
Open preview (PDF)Safety of machinery — Relationship with ISO 12100 — Part 5: Implications of artificial intelligence machine learning