Jinwoo Choi

Hi! I'm a Ph.D. student in the Department of Electrical and Computer Engineering at Seoul National University, advised by Prof. Seung-Woo Seo and affiliated with the Vehicle Intelligence Lab.

My research focuses on developing reinforcement learning algorithms for long-horizon decision-making. In particular, my goal is to enable agents to learn useful behavioral abstractions and make effective decisions toward distant goals. I also have experience in robotic exploration and autonomous-driving planning, including real-world testing with robotic and vehicle systems. Building on these experiences, I am interested in extending reinforcement learning methods to complex decision-making problems in robotics.

profile photo

Publications

FPAS exploration result FPAS: Frontier-Based Path Planning with Adaptive Sampling for Large-Scale Unknown Environments
IROS, 2026
clean-usnob Chain-of-Goals Hierarchical Policy for Long-Horizon Offline Goal-Conditioned RL
ICML, 2026
clean-usnob ESP-SLAM: Efficient Submap Partitioning for Large-Scale 3D Gaussian Splatting-Based SLAM
IEEE Robotics and Automation Letters (RA-L), 2026
clean-usnob Dynamic Contrastive Skill Learning with State-Transition Based Skill Clustering and Dynamic Length Adjustment
Jinwoo Choi, Seung-Woo Seo
ICLR, 2025
clean-usnob GIN: Graph-based interaction-aware constraint policy optimization for autonomous driving
IEEE Robotics and Automation Letters (RA-L), 2022
clean-usnob Ego‐lane index‐aware vehicular localisation using the DeepRoad Network for urban environments
IET Intelligent Transport Systems, 2021

Selected Projects

Hover over an image to play the video. On mobile, tap an image.

VISKY mobile robot VISKY: Mobile Robot for Autonomous Exploration in Unstructured Environments
Jan 2022 – Present

VISKY is a mobile robot developed for autonomous exploration in unknown and unstructured environments as part of an ADD-funded research project. I served as Project Manager for the overall project (January 2022 – June 2023) and led global path planning research (FPAS, IROS 2026). I also participated in real-world testing of the autonomous exploration system using a Clearpath Husky-based mobile robot.

SNUVI autonomous vehicle SNUVI: Autonomous Vehicle System for Urban Driving at K-City
Jul 2020 – Feb 2021

SNUVI is an autonomous vehicle developed for urban driving, integrating perception, localization, planning, and decision-making modules. As Team Leader of the overall project, I developed local path planning algorithms and participated in real-vehicle testing and system validation at K-City, a national autonomous driving testbed in South Korea.

Work Experience

Autonomous driving research vehicle ThorDrive
Research Intern | High-Level Decision-Making | Aug 2021 – Oct 2021

My work focused on hybrid high-level decision-making for autonomous driving, combining rule-based and reinforcement learning approaches for candidate-path selection in lane-change and detour scenarios. This work also involved constructing CARLA scenarios with illegally parked vehicles to investigate RL-based candidate-path selection.