Neural Rendering · Simulation · Generative Models
Junhyeok Lee
Making autonomous-driving simulation realistic enough to train and trust.
I'm a final-year undergraduate at Kookmin University, majoring in Automotive Engineering with a minor in Artificial Intelligence. Beyond autonomous driving, I'm interested in neural rendering and simulation, and in generative and world models for Physical AI. I will begin a corporate-sponsored master's program at Hyundai Motor Company, researching verification environments for autonomous driving.
Research interests
Neural rendering & simulationReconstructing drivable scenes from real sensor data, and building the simulators that make them usable for learning at training speed.
Generative modelsDiffusion-based domain adaptation and scene synthesis that narrow the Sim-to-Real gap, and metrics for how far an end-to-end model's evaluation in simulation can be trusted.
Experience
2026.01 — 2026.02
Research InternVehicle Intelligence Lab, Seoul National UniversityIntegrated datasets and trained learning-based perception models for autonomous navigation on a tracked off-road mobile robot, and optimized deployment on a Qualcomm NPU for real-time on-board inference.
2025.12 — 2026.01
Undergraduate Research InternKookmin UniversityStudied imitation learning and reinforcement learning for learning-based autonomous parking, and implemented and evaluated parking scenarios in the CARLA simulator.
2024.06 — 2025.11
Control TeamKUUVe · Kookmin University Unmanned VehicleLocalization and path tracking from odometry and GPS, and reinforcement learning with Sim-to-Real transfer on JetRacer.
Awards
2025.102025 International Robot Contest3rd place
2025.09HL FMA aMAP Innovator ChampionshipEncouragement
2025.05Autonomous Driving Robot Race2nd place
2024.12AutoRace, COSS3rd place
2024.102024 International Robot Contest2nd place