Mechanistic Interpretability for Learning Assurance of a Vision-Based Landing System

2026-05-19

AI Summary

Current approaches struggle to meet the European Union Aviation Safety Agency’s (EASA) evidentiary requirements for representation-level learning assurance in aviation AI systems. This work proposes the first context-aware representation analysis framework based on content/style disentanglement, employing a Vision Transformer to regress critical runway keypoints. By applying K-SVD sparse dictionary learning, patch embeddings are decomposed into interpretable atoms, and linear weight analysis validates a content-dominated decision mechanism. Building upon this, the authors introduce OOMS, a runtime assurance mechanism that effectively monitors the model’s operational scope, substantially enhancing existing out-of-distribution detection methods. This framework delivers the first verifiable, representation-level safety evidence for vision-based aircraft landing systems.

Abstract

EASA's learning-assurance guidance requires data-driven aviation systems to build and monitor their own situation representation, yet for neural networks the technical means to provide such evidence remain an open problem. We address this gap for a vision-based aircraft landing system: we propose that a minimally assurable model must at least be shown to separate content from style in its own situation representation. Showing that the model's predictions then rely largely on the contentful representation components leads to a concrete assurance path. To demonstrate this assurance path on a concrete model we train a vision transformer model for runway keypoint regression on the LARDv2 dataset. The model, which acts as the subject for our assurance demonstration, produces per-patch embeddings that we decompose into interpretable atoms via K-SVD sparse dictionary learning. A qualitative visualization confirms that contentful atoms track task-relevant runway structure and stylistic atoms track domain-specific appearance, and the regression head is shown to place almost all of its linear weight on contentful atoms. We further build on the content/style separation and define out-of-model-scope (OOMS) detection, a novel runtime assurance approach directly monitoring the model's situation representation. OOMS monitoring is complementary to operational design domain and output-space out-of-distribution monitoring and addresses concrete requirements of the recent EASA guidance. By directly analyzing a model's situation representation both at test time and runtime, this work delivers the first concrete piece of the representation-level evidence that EASA learning-assurance guidance demands, and points to mechanistic interpretability as a practical building block of future aviation safety cases.

Problem

• learning assurance
• situation representation
• mechanistic interpretability
• vision-based landing system
• EASA guidance

Innovation

• mechanistic interpretability
• content-style disentanglement
• learning assurance
• out-of-model-scope detection
• vision transformer