New algorithm brings GPS-level integrity to vision-based aircraft navigation

Vision-based navigation is a promising complement to Global Navigation Satellite Systems (GNSS), but certifying it for aviation requires strict integrity guarantees—assurances that faulty measurements won't cause dangerous errors. Previous work borrowed Receiver Autonomous Integrity Monitoring (RAIM) principles to detect faults in a probabilistic runway-based pose estimation pipeline. A new arXiv paper by Olivia Beyer Bruvik, Romeo Valentin, Marc R. Schlichting, Don Walker, and Mykel J. Kochenderfer extends that framework by deriving protection levels: probabilistic bounds on pose error that remain valid even when faults go undetected.

The authors present an algorithm that computes these protection levels for the nonlinear Perspective-n-Point (PnP) problem directly across all six degrees of freedom—both position and orientation. They analyze how measurement redundancy, pixel-level prediction uncertainty, and runway distance affect the resulting bounds, and demonstrate the tradeoffs on an illustrative runway approach. Accepted at the 2026 AIAA DATC/IEEE 45th Digital Avionics Systems Conference (DASC), this work provides a concrete path toward certifying vision-based systems for GPS-denied or degraded environments, potentially making them a reliable backup for commercial and general aviation.