Researchers at the University of Maryland’s A. James Clark School of Engineering have developed a groundbreaking vision-based control system for autonomous ship landings of Vertical Take-Off and Landing (VTOL) Unmanned Aerial Vehicles (UAVs). This innovation, led by Bochan Lee, Vishnu Saj, Moble Benedict, and Dileep Kalathil, addresses the critical challenge of landing UAVs on moving ships without relying on GPS signals.
The team’s solution draws inspiration from Navy helicopter landing procedures, where pilots use the ship as a visual reference for long-range tracking and a standardized visual cue called the “horizon bar” for the final approach and landing phases. The researchers have automated this process using a uniquely designed nonlinear controller integrated with machine vision. The vision system employs machine learning-based object detection for long-range ship tracking and classical computer vision techniques to estimate the aircraft’s relative position and orientation using the horizon bar during the final approach and landing phases.
The nonlinear controller operates based on the information provided by the vision system and has demonstrated robust tracking performance even in the presence of uncertainties. This innovative approach ensures precise and safe landings, which are crucial for maritime operations. The developed autonomous ship landing system was successfully implemented on a quad-rotor UAV equipped with an onboard camera. The researchers conducted extensive simulations and flight tests to validate the system’s vertical landing safety, tracking capability, and landing accuracy. The successful demonstration of approach and landing on a moving deck, which mimics realistic ship deck motions, underscores the system’s potential for real-world applications.
This research represents a significant advancement in the field of autonomous maritime operations. By eliminating the need for GPS signals, the system enhances the reliability and safety of UAV landings on ships, which is particularly important in GPS-denied environments. The integration of machine learning and classical computer vision techniques provides a robust and adaptable solution that can be applied to various maritime scenarios. The successful implementation and testing of this system pave the way for broader adoption of autonomous UAVs in naval and commercial shipping operations, ultimately improving efficiency and reducing the risk of accidents. Read the original research paper here.