Keimyung University Robotics Engineering Department Wins Top Prize at the "2026 Marine Engineering CAE Contest"

A Technique to Improve the Safety of Autonomous Ships at Night
Developed Thermal Imaging-Based AI Technology to Detect Other Vessels
Generative AI Overcomes Data Shortages
Research Presented at Korean Society of Ocean Engineers Fall Conference

A student team from the Department of Robotics Engineering at Keimyung University (President Shin Ilhee) developed AI-based technology to detect other vessels, helping autonomous ships navigate safely at night, and won the top prize at the 2026 Marine Engineering CAE Contest.

The awards ceremony was held on October 15 at Exhibition Center 2 at BEXCO in Busan, where the winning team received a certificate and 1 million won in prize money.

The student team from the Department of Robotics Engineering at Keimyung University poses for a commemorative photo after winning the top prize at the awards ceremony for the 2026 Marine Engineering CAE Competition, held on the 15th at BEXCO in Busan. Photo provided by Keimyung University.

The team consisted of third-year students Lee Hyeongjin, Kim Jaeyun, and Son Yeowon from Keimyung University's Department of Robotics Engineering, under the guidance of Professor Hong Sunghun.

They presented their research on "A Thermal Imaging-Based Automatic Detection Method for Other Vessels to Ensure the Safe Navigation of Autonomous Ships in Nighttime Maritime Environments."

Organized by the Korean Society of Ocean Engineers and sponsored by the Ministry of Oceans and Fisheries and the Korea Offshore & Shipbuilding Association, among others, the contest is a nationwide undergraduate competition featuring open topics in marine engineering, including ships and offshore platforms, and marine robotics.

For the final round, top teams from 10 universities across the country, including Keimyung University, KAIST, Seoul National University, and Pusan National University, advanced to compete.

The winning team focused on the limitations of ordinary RGB cameras, which struggle to recognize nearby vessels in low-light conditions at night or in severe weather.

Thermal infrared (TIR) cameras, considered an alternative, are less affected by changes in illumination. However, a major challenge was the severe shortage of maritime thermal imaging data for training AI.

To address this challenge, the students devised a technology that automatically converts existing maritime RGB footage into thermal images.

In particular, they applied a content-preserving generative AI model to prevent changes to vessels' positions and shapes, while retaining the location information in the original footage.

The project was highly regarded for reducing the laborious need for people to label images manually and for efficiently building large-scale training datasets.

After winning the top prize, the team also attended the fall academic conference marking the 40th anniversary of the Korean Society of Ocean Engineers and presented their research findings to the academic community, attracting attention.

Student Lee Hyeongjin said, "We continued our research by carefully considering ways for autonomous ships to reliably recognize nearby vessels in nighttime maritime environments. We will use this award as a springboard to further develop our research capabilities so that we can apply AI and robotics technologies in real-world maritime settings."

Professor Hong Sunghun said, "It is especially meaningful that the students independently identified real technical challenges in the maritime sector and proposed practical solutions using generative AI. We will continue to strengthen students' problem-solving skills through practice-oriented education and research that integrates robotics and AI."

This achievement demonstrates the potential of generative AI to address technical challenges in the maritime sector and is expected to contribute to advancing technologies for the safe navigation of autonomous ships.

This content was produced with the assistance of AI translation services.