• Journal of the Korea Institute of Military Science and Technology
• /
• v.27 no.6
• /
• pp.677-683
• /
• 2024

Landing a helicopter on a moving ship requires accounting for the ship's attitude. However, it is not only challenging to visually assess the ship's attitude, but it also creates illusions for the pilot, increasing the risk of accidents. In this study, we propose an image-based ship attitude estimation method to assist helicopter landings. The proposed method enhances landing safety by predicting the ship's heave, pitch, and roll using only helicopter-mounted optical devices and pre-trained deep learning models, without requiring communication with the ship. To implement this approach, we generated a dataset by simulating a virtual sea environment and ship motion. Using this data, we trained deep learning models to predict the ship's attitude based solely on images. Experimental results confirm the feasibility of the proposed method, with VGG-16 demonstrating particularly effective attitude prediction under simulated conditions.