Evidence & review
From raw evidence
to an explainable result.
RaceAmate's verification engine is our own work in progress. It exists to answer one question honestly: what does the evidence actually show?
The pipeline
Observe. Measure. Bind. Re-check.
01 / Observe
Original video
The original recording of the actual visible green is kept, with its frame timing and exposure information. RaceAmate observes the light. It does not predict it.
In development02 / Measure
Raw Dragy telemetry
Raw receiver data from a compatible Dragy unit (speed, position and fix quality) is retained, and the video clock is related to the receiver clock.
In development03 / Bind
The right run
Evidence is tied to the actual car, driver, devices, event, course and ruleset. Interruptions and uncertainty are preserved, not smoothed away.
In development04 / Re-check
Server review
The server rebuilds the run from the evidence itself, re-runs the checks and records the software version. Disagreement is explained, not hidden.
In development
The phone submits evidence. It does not award an official result.
What counts as evidence
Raw evidence, not summaries.
A number typed in or shown on a screen is a claim. Evidence is the raw material that lets someone else check it.
Original video
The original recording of the actual visible green, with frame timing and exposure information. Not a re-encoded or trimmed copy.
Raw Dragy telemetry
The receiver's raw speed, position and fix-quality data, not only its summary times.
Calibrated bindings
Clock alignment between video and receiver, and the mapping of the real course, each with its error bounds.
AI can help find candidate frames and raise flags. Reproducible measurements and agreed rules decide eligibility.
The start
Same start, same finish.
- Rolling entryCars A and B roll side by side at the organiser's rolling speed.
- Actual greenThe real, human-controlled green light marks the start. It is observed, never predicted.
- Common finishBoth cars race to the same physical finish line.
- Venue format varies. The real course must be validated.
A personal distance from the green is not a shared finish. Course mapping, clock alignment and error bounds must be validated. A finish order that can't be resolved stays unresolved.
Outcomes
Eligible, inconclusive, or not eligible.
Eligible
The evidence is complete, consistent and within the agreed rules. Only eligible performance can set or raise a class.
Inconclusive · review required
Something is missing, uncertain or disagrees. It stays unresolved and goes to a person. It is never guessed into a result. Drivers see “Under review”.
Not eligible
The evidence shows the run cannot count, for example a start outside the organiser's tolerance. A ruling always carries a written reason.
People, not just software
How review works.
Engine doubt goes to review
When the engine flags a submission, it is queued for a moderator. The driver sees “Under review”, never engine jargon or an accusation.
Rulings carry reasons
Every ruling records a written reason. It is the one record the driver always gets to see.
Disputes are separate
A person challenging a published run is a different process from the engine's own doubt, and they are kept apart.
Officials decide the event
On the night, the organiser's officials make the calls. RaceAmate's review is about what the evidence shows, afterwards.
Limits, stated plainly
What it is not.
Not validated for automatic winners or live officiating
Not perfect, and not instant
Earlier records