LARD 2.0: Enhanced Datasets and Benchmarking for Autonomous Landing Systems

2026-03-23

AI Summary

This study addresses the limitations of insufficient training data diversity and unrealistic operational design domains (ODDs) in autonomous landing systems by integrating BingMap aerial imagery with flight simulator data to expand the LARD dataset. The resulting LARD 2.0 encompasses a more realistic ODD covering multiple-runway airports and introduces a novel object detection benchmark tailored for complex, multi-instance scenarios typical of real-world airfields. This work presents the first effective fusion of real-world and synthetic data, significantly enhancing model generalization. To support further research, the authors release the LARD 2.0 dataset, baseline models, and evaluation tools as open-source resources, establishing the first dedicated object detection benchmark for autonomous landing tasks. Empirical results demonstrate superior detection performance and robustness in challenging airport environments.

Abstract

This paper addresses key challenges in the development of autonomous landing systems, focusing on dataset limitations for supervised training of Machine Learning (ML) models for object detection. Our main contributions include: (1) Enhancing dataset diversity, by advocating for the inclusion of new sources such as BingMap aerial images and Flight Simulator, to widen the generation scope of an existing dataset generator used to produce the dataset LARD; (2) Refining the Operational Design Domain (ODD), addressing issues like unrealistic landing scenarios and expanding coverage to multi-runway airports; (3) Benchmarking ML models for autonomous landing systems, introducing a framework for evaluating object detection subtask in a complex multi-instances setting, and providing associated open-source models as a baseline for AI models' performance.

Problem

• autonomous landing systems
• dataset limitations
• object detection
• Operational Design Domain
• benchmarking

Innovation

• autonomous landing
• dataset enhancement
• Operational Design Domain (ODD)
• object detection benchmarking
• multi-runway airports