Vision-Based Agile Landing on Turbulent Waters

2026-05-22

AI Summary

This work addresses the challenge of enabling autonomous agile landing of unmanned multirotor aerial vehicles on moving ships under high-sea conditions without access to explicit platform state information. The authors propose a reinforcement learning approach that fuses the vehicle’s own state with local visual features—specifically keypoints and their descriptors—of the landing surface to directly output attitude and thrust commands, which are then executed by a low-level controller. A key innovation lies in achieving turbulent-sea landings without relying on explicit platform state estimation for the first time. The method further introduces a training strategy based on synthetic keypoints and randomly normalized descriptors, facilitating zero-shot transfer to different onboard feature extractors. Simulations demonstrate superior performance over state-of-the-art model predictive control baselines, and real-world flight experiments validate reliable landing capability across two distinct visual feature extractors.

Abstract

Autonomous landing of Unmanned Aerial Vehicles on maritime vessels is challenging due to the coupled motion of the vehicle and landing platform in open-sea conditions. This paper presents a reinforcement-learning-based approach for autonomous multirotor landing on moving maritime platforms without requiring explicit platform-state information. The proposed method uses multirotor state measurements together with local visual features, consisting of keypoints and associated descriptors extracted from the landing surface, to predict attitude and thrust commands. These commands are tracked by a conventional low-level controller. The policy is trained in simulation using synthetic keypoints with randomly generated normalized descriptors, enabling zero-shot deployment with different local feature extractors onboard the UAV. We evaluate the method in a realistic simulator and show that it outperforms a state-of-the-art Model Predictive Control baseline under platform motions corresponding to ``Very Rough'' sea conditions. Finally, we perform extensive real-world experiments, demonstrating autonomous onboard landing using two different local feature extractors. To the best of our knowledge, this is the first approach for agile multirotor landing on maritime platforms in turbulent waters that does not rely on an explicit platform-state representation.

Problem

• autonomous landing
• maritime platforms
• turbulent waters
• Unmanned Aerial Vehicles
• agile landing

Innovation

• reinforcement learning
• vision-based landing
• maritime UAV
• zero-shot transfer
• local features