Chun-Kit Li, Iok Long Sit, Ming Fung Siu, Ka Yu Kui, Hin Wang Lin, Pengyu Wang, Ling Shi
Cheng Kar-Shun Robotics Institute
The Hong Kong University of Science and Technology
WaveLander is a reinforcement-learning-based hierarchical control framework for UAV landing on wave-disturbed platforms. It decouples high-level vertical landing decision-making from low-level flight stabilization, enabling timing-aware descent, holding, and retreat behaviors under dynamic platform motion.
🚧 Code release coming soon.
This paper has been submitted to ICARCV 2026 and is currently under review.
The repository is being prepared for public release.
WaveLander is a hierarchical learning-based control framework for UAV landing on wave-disturbed marine platforms. Instead of learning low-level motor commands, WaveLander uses reinforcement learning for high-level vertical landing decisions, while a conventional flight controller handles attitude stabilization, lateral tracking, and velocity tracking.
The learned policy takes a compact platform-relative observation, including relative height, vertical velocity, platform tilt, and tilt-rate information, and outputs a scalar vertical velocity reference. This formulation reduces dynamic platform landing to a low-dimensional timing-aware control problem, where the UAV learns when to descend, hold, or retreat under time-varying platform motion.
WaveLander is evaluated through randomized MuJoCo simulation, Isaac Sim software-in-the-loop transfer, and a representative real-world deployment test. The results show that the learned vertical policy improves touchdown timing compared with fixed-descent behavior and provides a compact interface for deployment-oriented UAV landing on moving marine platforms.
The manuscript is currently under review.
Paper link and citation information will be updated after publication.
Supplementary experiment videos will be added soon.
Citation information will be updated after the paper becomes available.
@misc{li2026wavelander,
title = {{WaveLander}: A Generalizable Hierarchical Control Framework for {UAV} Landing on Wave-Disturbed Platforms via Reinforcement Learning},
author = {Li, Chun-Kit and Sit, Iok Long and Siu, Ming Fung and Kui, Ka Yu and Lin, Hin Wang and Wang, Pengyu and Shi, Ling},
year = {2026},
note = {Submitted to ICARCV 2026, under review}
}We thank Prof. Ling Shi for his guidance and support throughout this research project. We also thank HKUST and the Cheng Kar-Shun Robotics Institute for their support. We are grateful to our colleagues and close collaborators at the 3121B Multi-Agent Systems Laboratory, including Fan Zhang, Yim Ying Hing, Shan Wen, and Xiawei Du, as well as Shiliang Zhao from Zhengzhou University, for their helpful support during the preparation of this work.
The code will be released under the MIT License upon public release.