operational design domain
In plain wordsThe specific conditions an automated vehicle or robot is designed to work in, such as particular roads, speeds, weather and times of day, outside of which it must not operate on its own.
The envelope of conditions that an automated driving system, or more generally any autonomous system, is engineered to operate within: geographic area and road types, speed range, weather, lighting and time of day, traffic conditions, and required infrastructure such as lane markings or connectivity. The ODD bounds a system's safety claims and is central to its validation, approval and marketing: a robotaxi may be validated only for certain city districts in good weather, a highway system only for divided motorways below a set speed. When a system detects that it is leaving, or about to leave, its ODD, it must either hand control back to a human (Level 3levels of driving automation, n.: A six-step scale, from Level 0 to Level 5, describing how much of the driving a car's automation does and how much the human must still watch the road and be ready to take over.) or reach a minimal risk condition by itself, such as stopping safely (Levels 4–5); only Level 5 implies an unlimited ODD. The concept has spread to drones, mobile robots and AI systems generally as a way of stating the conditions under which performance claims hold.
Heavy snow is outside our ODD, so the fleet returns to the depots when the forecast turns.
Origin
Defined in SAE International's J3016 taxonomy of driving automation and adopted in US federal automated-vehicle policy from 2016 and in international regulation such as UN Regulation No. 157 on automated lane keeping (in force January 2021). Standards such as BSI PAS 1883 (2020) and ISO 34503 (2023) specify how to describe an ODD.
Usage note
Related to, but more operational than, a machine-learning model's training distribution: inputs that are out-of-distributionout-of-distribution, adj.: Describes inputs that differ from anything an AI system saw in training — like a model trained on daytime photos facing a night scene — where its answers become less reliable. for a perception model can still occur inside a nominal ODD.
Related
- levels of driving automationlevels of driving automation, n.: A six-step scale, from Level 0 to Level 5, describing how much of the driving a car's automation does and how much the human must still watch the road and be ready to take over.
- autonomous vehicleautonomous vehicle, n.: A vehicle, usually a car, truck or shuttle, that can drive itself, sensing the road, deciding what to do and steering, braking and accelerating with no human driving.
- robotaxirobotaxi, n.: A taxi with no human driver: you summon it with a phone app, and a self-driving car picks you up and drives you to your destination.
- advanced driver-assistance systemadvanced driver-assistance system, n.: Car features that help the driver but do not replace them, such as automatic emergency braking, lane-keeping help, adaptive cruise control and blind-spot warnings.
- out-of-distributionout-of-distribution, adj.: Describes inputs that differ from anything an AI system saw in training — like a model trained on daytime photos facing a night scene — where its answers become less reliable.
- safety casesafety case, n.: A written, evidence-backed argument that a particular AI system is safe enough for a particular use, putting the burden of proof on the developer, as with nuclear plants or aircraft.
Sources
- BSI (2020). PAS 1883: Operational design domain (ODD) taxonomy for an automated driving system (ADS) — Specification. British Standards Institution.
Defined in law or standards
SAE J3016_202104 (definition of 'operational design domain')
Mentioned in
autonomous vehicleautonomous vehicle, n.: A vehicle, usually a car, truck or shuttle, that can drive itself, sensing the road, deciding what to do and steering, braking and accelerating with no human driving., levels of driving automationlevels of driving automation, n.: A six-step scale, from Level 0 to Level 5, describing how much of the driving a car's automation does and how much the human must still watch the road and be ready to take over., robotaxirobotaxi, n.: A taxi with no human driver: you summon it with a phone app, and a self-driving car picks you up and drives you to your destination.