Heimdall technical evidence room
← Public site

// ARCHITECTURE + VALIDATION REGISTRY

Defined build.
Unproven performance.

The near-Nairne pilot is in development. Property permission is secured and the hardware, evidence and operating foundations are defined; no complete Heimdall unit is assembled and field work has not begun. This page records what the build is intended to do, what remains to be measured and how funding is allocated without overlap.

// FIVE-STEP OBSERVATION CASCADE

  1. 01

    Wake

  2. 02

    Gate

  3. 03

    Identify

  4. 04

    Hand off

  5. 05

    Review

Motion wakes the system; thermal sensing gates camera use; RGB/IR captures reviewable evidence; long-range radio passes the event; a human reviews the timestamp, node or node-pair, heading, confidence tier and raw thumbnail. Heimdall does not dispatch.

// PHYSICAL BUILD SEQUENCE

Commodity first. Custom when earned.

The MVP uses verified off-the-shelf weatherproof enclosures, cable glands and mounting hardware. A custom housing and production tooling are not required to test the sensing cascade, radio path, power budget or field-placement hypothesis.

01 / PROVE

COTS enclosure

Assemble the minimum hardware needed to retire compatibility, weather and field risks. The US$1,000 target does not fund a printer or production tooling.

02 / ITERATE

Print selectively

Demonstrated interest and iteration demand may justify an in-house printer for adapters, internal carriers, hoods, mounts and assembly jigs.

03 / VALIDATE

Choose production

Outdoor parts still need ingress, UV, heat, condensation and mechanical testing. Volume economics will determine printing, contract production or moulding.

// WHAT THE TRIAL MUST PROVE

Measurement gates

  • Maximum reliable identification distance in real night, terrain and weather conditions
  • Fox-versus-dog confusion rates using separate held-out field evidence
  • Fence-post-height long-range radio performance and relay queue behaviour
  • Power use, battery/solar autonomy and weather resilience

// NAIRNE MVP FUNDING REGISTRY

US$1,000 total target

The target funds the current near-Nairne deployment plan as a whole. Per-unit amounts are assigned only from current quotes; the site does not invent a split. Every accepted allocation is recorded once against a stable ID.

RECORDED STATUS
US$0 / US$1,000

US$1,000 remaining · 0% allocated

HMD-NRN-CORE-01

Core BOM and hub

Scope
Sensing, compute, radio, power and mounting for the defined core deployment.
Measurement purpose
Establish the complete first near-Nairne MVP deployment.
Evidence state
planned
Allocation state
open
Funder
—
HMD-NRN-INT-01

Bench integration

Scope
Part compatibility, firmware integration and a recorded repeatable build.
Measurement purpose
Retire integration risk before field installation.
Evidence state
planned
Allocation state
open
Funder
—
HMD-NRN-EXP-01

Corridor nodes

Scope
Additional sensing points within the first deployment where the final BOM permits.
Measurement purpose
Measure incremental coverage and node-to-node handoff.
Evidence state
planned
Allocation state
open
Funder
—
HMD-NRN-DATA-01

Training and validation

Scope
Human review of field imagery with training observations separated from held-out validation.
Measurement purpose
Produce honest fox-versus-dog evidence without reusing evaluation data.
Evidence state
planned
Allocation state
open
Funder
—

Not open. A second-trial target will be considered only after the near-Nairne target is fully allocated.

Safety and privacy boundaries

A human always reviews and decides. No autonomous dispatch.
No face recognition, named biometrics or intent labels.
Every classification carries a confidence tier and raw thumbnail.