Yujin Kim

Yujin Kim

Cornell University



My research interest lies in the area of Reinforcement Learning, both practical applications and theoretical perspectives. I focus on solving sequential decision-making problems for real-world autonomous learning systems! These are the questions that I am digging these days.

1️⃣ How can an agent acquire skills with minimal (or without) human-engineered intervention?

2️⃣ How can an agent acquire general-purpose skills to efficiently solve practical, long-horizon tasks?

3️⃣ What are the key differences that distinguish AI agents from humans in the process of learning new skills? What aspects of human learning can AI systems leverage? I am a Ph.D. student in Computer Science at Cornell University, working with Sanjiban Choudhury. Before joining Cornell, I was a research scientist at the Korea Institute of Science and Technology (KIST). I received my M.S. from Korea University and my B.S. from the University of Seoul.

📖 Educations

  • Ph.D. in Computer Science, Aug. 2024 - Present
  • M.S. in Electrical and Computer Engineering, Mar. 2021 - Aug. 2023
    • Major : Control, Robotics and Systems. Korea University (GPA: 4.39/4.5)
    • Research Scholarship from Hyundai Motor Group, Mar. 2021 - Dec. 2022,
  • B.S. in Electrical and Computer Engineering, Mar. 2017 - Feb. 2021
    • University of Seoul. (GPA: 4.0/4.5)
    • Research Scholarship from Hyundai Motor Group, Sep. 2019 - Dec. 2020

📝 Publications

sym

Distilling Realizable Students from Unrealizable Teachers

  • Policy distillation under asymmetric imitation learning setting.
  • Propose two new IL/RL algorithms robust to state aliasing.
  • Submitted to IROS 2025
sym

[Autonomous Robot Manipulator Operation for Intricate Object Handling] (https://arxiv.org/abs/2412.08522)

  • Develop skills for operating equipment (e.g., valve, switch, gear lever…) at industrial sites with a manipulator.
  • Learn skills while minimizing human-engineered features using reinforcement learning.
sym

Dual-Armed Mobile Manipulator Door Traversal

  • Address challenges of door traversal motion planning
  • Unified framework for door traversal, from approaching, opening, passing through, and closing the door with dual-armed mobile manipulator
  • Decision making for optimal contact point planning with RL.
sym

Classified Experience Replay

  • Challenge to skewed sub-goal distribution for goal-conditioned RL controller.
  • Enable adaptive sub-goal planning and efficient reward learning via MPC-synchronized rewards.
sym

Learning to drive in highway with guided RL controller.

  • Addresses the challenge of reward shaping for continuous RL controllers by using MPC reference. <!– - Implemented DDPG.
  • Paper published at ICEIC, 2022 (oral). –>

🎖 Honors and Awards

  • 2024 Student Travel Grant, ICRA 2024 (MOMA.v2 Workshop)
  • Spring 2018, Scholarship for Excellent Achievement, University Of Seoul.
  • Sep. 2018 - Dec. 2022, Full Scholarship for Selected Research Student, Hyundai Motor Company.
  • May 2022, 10th F1TENTH Autonomous Racing Grand Prix, 3rd Place, ICRA 2022.
  • Jul. 2018, 2018 Intelligent Model Car Competition, 3rd Place, Hanyang University.
  • Jul. 2017, 14th Microrobot Competition, Special Award for Women Engineer, Dankook University.

💻 Work Experience