Our university's automotive AI club ‘KaAI’ achieved 2nd place in the Future Trajectory Prediction category at the 2025 Autonomous Driving AI Challenge hosted by the Ministry of Science and ICT. At the awards ceremony held on November 14, they accomplished the great feat of receiving the Director General of the Institute for Information & Communications Technology Planning & Evaluation Award (prize money: 2.5 million won).

The Autonomous Driving AI Challenge is a nationwide technical competition evaluating the perception, judgment, and prediction capabilities of autonomous vehicles based on real-world driving data. Held annually, it brings together universities, research institutes, and companies to chart the course for autonomous driving AI technology advancement. This year's challenges comprised three advanced categories demanding heightened cognitive and predictive abilities compared to last year's theme: ① 3D Object Detection, ② Semantic Segmentation, ③ Future trajectory prediction. These three areas, selected to demand more advanced perception and prediction capabilities than last year's topics, were successfully conducted with 154 teams participating for a total prize pool of 22.5 million won (awarded to 1st, 2nd, and 3rd place teams per category).

KaAI undergraduate team members Kang Sang Gu, Seo Eun, Yang Jeong Woo, and Jeong Seong Hyeon submitted “ETDeMo,” a trajectory prediction model that forecasts a vehicle's future path in diverse driving environments, for the third category. This newly developed model was meticulously designed to simultaneously consider the global scene context, vehicle-to-vehicle interactions, temporal consistency, and driving intent. It demonstrated stable prediction capabilities even in complex situations such as intersections, curves, and forks. Notably, it maintained high precision in these complex environments and recorded excellent performance across all evaluation metrics.

This competition was conducted using a public dataset built through the Autonomous Driving Technology Development Innovation Project, focusing on the technical challenge of how accurately autonomous vehicles can predict future driving situations. Among leading research teams nationwide, the KaAI team achieved second place, recognized for the model's completeness and practical applicability.

Professor Lee Sang Hun, the team's advisor, stated, “Our undergraduate team securing second place in a government-hosted competition where most participants are AI graduate students is a triumph that proves the excellence of our university's automotive education program.”

KaAI is a research-focused club centered around undergraduate students from Kookmin University's  College of Automotive Engineering, conducting research and projects in autonomous driving, computer vision, and artificial intelligence. The club continues its active engagement, consistently achieving outstanding results in domestic and international competitions and academic events. Recent accomplishments include winning first place in the IEEE Intelligent Transportation Systems Society (ITSS) AI competition, placing third in the CVPR 2025 DriveX Grand Challenge, and being selected for an oral presentation at IEEE IV 2025.