DGIST’s Seonghun Yoon and MIT’s Hyungtae Lim Joint Research Team Wins First Place at World’s Leading Robotics Conference
It was announced on June 9 that the joint research team led by Professor Seonghun Yoon from the Department of Electrical Engineering and Computer Science at DGIST and Postdoctoral Researcher Hyungtae Lim from MIT achieved first place among 56 teams worldwide in the 'GOOSE 2D Semantic Segmentation Challenge,' held at the Field Robotics Workshop of the 2026 ICRA (International Conference on Robotics and Automation), the world’s most prestigious robotics conference.
This achievement is significant not only because it demonstrates the excellence of South Korea’s artificial intelligence (AI)-based image recognition technology on the global stage, but also because it highlights the scalability of core perception technologies essential for implementing 'Physical AI' that interacts with real-world environments.
The joint research team participants (from left) are Seonghoon Yoon, Sangjin Lee, Hyobin Choi, Jaeil Park from DGIST, and Hyungtae Lim from MIT.
View original imageThe challenge, co-hosted by the Fraunhofer IOSB Institute in Germany, Bundeswehr University Munich, and the University of Koblenz, evaluated how precisely field robots can understand complex scenes in unrefined real-world environments. Unlike conventional autonomous driving datasets collected primarily on well-structured urban roads, the 'GOOSE dataset' used in this competition is characterized by data centered on field robots operating in unpredictable and unstructured outdoor settings.
The dataset used in this competition consisted of unstructured outdoor data collected from various platforms, such as excavators and quadruped robots, making it more demanding than typical urban road environments. Notably, this year, the number of evaluation categories expanded dramatically to 64 detailed classes, requiring a sophisticated ability to accurately recognize 'long-tailed classes'—rare objects that appear extremely infrequently in the field.
The research team independently developed a framework that organically combines Meta’s latest self-supervised foundation model, 'DINOv3,' with the image segmentation model 'Mask2Former.' This system maintained robust visual recognition performance despite numerous variables that can arise in real outdoor environments, such as changes in lighting, irregular terrain, and complex backgrounds.
In particular, by maximizing the ability to detect rare objects that AI models often miss due to data scarcity, the system dramatically reduced catastrophic failures—critical recognition errors that could directly lead to accidents—thereby enhancing safety. As a result, this breakthrough is expected to enable the broad expansion of field robot technologies into a wide range of industrial sectors, including autonomous vehicles, disaster response, smart agriculture, and construction sites.
Professor Seonghun Yoon of DGIST’s Department of Electrical Engineering and Computer Science said, “The technology to precisely understand scenes in unpredictable and unstructured outdoor environments is the most essential for ensuring the autonomy and safety of field robots. Building on this success on the global stage, we will continue to advance research on powerful visual perception technologies that can be immediately applied to real robot systems and various industrial sites.”
Hot Picks Today
"Two Spend 500,000 Won in a Flash, Yet Wait Two Hours?" ... 'Dreamland' Shunned by Kids and Foreign Tourists
- The Era of Getting 200 Million Won After 30 Years Is Over... A Japanese Bank's "Perpetual Department Head" Who Just Clocked In Cries Out
- Spain Hit by Its Worst Tragedy... Lee Kangin, Moreno and Real Madrid Join Soccer World in Mourning
- They Said, “Make It Your Home for Life” - Then Suddenly Kicked 200 Passengers Off... What Happened to the Around-the-World Cruise?
- "130 Rice Bowls Sold Out in a Day"... 'Unfiltered Face' Adds 8 Million Won to Night Market Sales: China's Bungeunwoo
Meanwhile, this accomplishment is the result of close global cooperation between DGIST and leading overseas research institutions, and it is drawing attention as an important stepping stone for DGIST to become a central hub for global robotics and artificial intelligence research.
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.