Dry highway, clear, low density
Unremarkable
A firm brake at highway speed in clear conditions is routine — context says so.
Platform · ODD Engine
Read risk in its true context.
Raw telemetry is meaningless without context. The ODD Engine tags every machine event with the conditions it happened in — so AURA scores behaviour on the ground, not in the abstract. AURA supports underwriting; the licensed insurer retains pricing authority.
Why context matters
The same move, two different stories.
A hard brake is unremarkable in one setting and alarming in another. The ODD Engine attaches operational-design-domain tags at ingest, so scoring and insurer review see the same facts.
Wet downgrade, dusk, school zone
Elevated
The identical deceleration in this context is a meaningful risk signal.
What we annotate
Every event inherits a context vector.
Each trip event carries a structured set of tags used by the scoring model and the audit trail.
Road & route
Road class, lanes, surface, curvature, and route adherence.
Speed regime
Speed limit, speed band, and how the machine sits within it.
Gradient & geometry
Slope, elevation change, and the shape of the road ahead.
Weather bands
Rainfall, temperature, humidity, and UV at the time of travel.
Urban context
Density, road environment, and proximity to sensitive zones.
Traffic density
Surrounding flow, so behaviour is read against real conditions.
Compounding exposure
Annotated history is a moat.
A newcomer cannot retroactively label conditions it never recorded. YAS accumulates annotated kilometres with every operator trip — the context library deepens continuously, and that depth is what makes scores defensible in an insurer review.
- Context tags are written at ingest, not reconstructed later
- Every operator trip widens the annotated baseline
- Depth of context is what makes a score explainable and auditable
Downstream scoring
Context-weighted, explainable factors.
Annotated events feed the roughly 100-factor scoring model. Context-weighted factors — environment-specific speeding, grade-adjusted braking, weather-conditioned frequency — move the AURA score while keeping every contribution traceable to a behaviour in a setting.
Environment-specific speeding
Speeding is judged against the limit and conditions that actually applied.
Grade-adjusted braking
Braking on a downgrade is weighed differently from braking on the flat.
Weather-conditioned frequency
Event frequency is read against the weather it occurred in.
See how it fits the platform.
The ODD Engine is one layer of the YAS pipeline — capture, context, scoring, attestation. See the whole picture.