Benchmark

Benchmark

This page frames fair comparison by navigation problem class — long-horizon, globally referenced position when GNSS and operator radio for navigation are denied — not competitor attack slides. Illustrative figures and measured field results are labeled separately.

Comparison frame

Compares

  • VIO, SLAM, IMU dead-reckoning, GNSS, and VIN as navigation classes under GNSS- and radio-denied assumptions.
  • Failure modes on long-horizon missions: unbounded local drift vs map-backed global state.
  • Whether a method maintains a persistent global reference without satellites or an operator nav link.

Does not compare here

  • Named competitor marketing claims or bake-offs without a shared method.
  • Local obstacle avoidance, mission AI, or perception stacks that are not global navigation.
  • Price, BOM, or brand positioning.

Illustrative class comparison

Illustrative

Illustrative signed cross-track deviation: VIN near zero; VIO and SLAM stay close then diverge slightly; IMU error is worst (GNSS and datalink denied).

Illustrative deviation from the reference track (meters). VIN in the shaded corridor; generic VIO and SLAM track together then separate slightly; IMU dead-reckoning shows the largest error under GNSS and operator-datalink denial. This chart is the same public figure used on Visual Autonomy — it shows class behavior, not a substitute for measured field scoring.

Same chart on Visual Autonomy → · GNSS- and radio-denied framing →

Measured basis in this portal

Field scoring of VIN against a GNSS reference is published separately — method, sites, and scored sorties — not restated as a bake-off table here.

  • Primary measured metrics: locked-mode horizontal error (median, mean, max) and vision lock duty-cycle.
  • Current field sites: Ray Brickhouse Flying Field and Scobee Flying Field (AMA community, Texas).
  • Full entries, video, and Excel reports: Flight Log. Snapshot and method: Independent Validation.

Fairness notes

  • Same problem class and denial assumptions before any figure is read as a comparison.
  • Illustrative charts are labeled Illustrative; measured sorties use the Validation / Flight Log method.
  • Peer products often publish different metrics (local drift, obstacle performance, mission autonomy). Apples-to-apples comparison here is only on globally referenced navigation under denial.
  • Independent third-party witness campaigns are not part of the present measured set; they will be reported when complete.

What this page describes — and what it does not

Describes

  • Export-safe class framing for VIN vs VIO / SLAM / IMU / GNSS under denied conditions.
  • The public illustrative navigation-error chart, reused here with the same Illustrative label.
  • Pointers to measured field validation already published in this Briefing Center.

Does not include

  • A full measured competitor pack — staged for L1 access when ready.
  • Qualitative failure-mode table — Limitations of Existing Navigation.
  • Architecture depth — Technology / vault materials under L1 access.

Independent Validation → · Limitations → · Technology → · Visual Autonomy (denied nav) →