Student Team from the School of Information, Communication and Electronic Engineering Wins Top Prize at the ‘5th Future Mobility Industry Idea Contest’
- Writer :External Affairs Team
- Date :2026.10.06

A team of students from the School of Information, Communications & Electronics Engineering at The Catholic University of Korea—Oh Yeon-ju, Won Chi-hun, and Kim Do-kyun—won the Grand Prize (K-Mobility Bridge Foundation Chairman’s Award) at the "5th Future Mobility Industry Idea Contest" on September 16. The three students are alumni of the Wireless Communication Research Laboratory led by Professor Emeritus Choi Sang-ho and currently work as researchers at S&Y Corporation, a company founded by Professor Choi.
Hosted by the K-Mobility Bridge Foundation and sponsored by the Ministry of Trade, Industry and Energy and Hyundai Motor Group, the contest was organized to discover innovative technology ideas and talent capable of responding to the paradigm shift in future mobility. This year, seven teams—selected through a review of research proposals and papers across the fields of mobility, AI, robotics, and ESG—advanced to the final presentation round.
The winning team proposed a method titled "Selective Imitation Learning for Autonomous Buses via Class-Balanced Data Curation." Their approach involved using data from situations requiring manual intervention during autonomous driving to intensively reinforce the system's weak points.
To achieve this, they applied a "class-balancing" technique to maintain a balance of data across various scenarios—such as sudden stops, lane changes, and driving on curved sections—thereby improving weaknesses while preserving existing driving performance. Validation using the autonomous driving simulator CARLA confirmed improvements across three key metrics: driving stability, ride comfort, and the maintenance of existing performance levels.
This research demonstrated that the performance of autonomous driving models can be effectively enhanced through data curation—the process of selectively curating and organizing training data to suit specific objectives. In particular, the use of data on manual interventions—captured during actual operation—for machine learning is expected to contribute to the future advancement of autonomous driving software.

● Text/Photos: External Relations Team, Intern Seong Hyeon-seo

