Hojune Kim

Hojune Kim

I am a Master's student in Aeronautics and Astronautics at Stanford University, advised by Prof. Mac Schwager in the Multi-Robot Systems Lab (MSL). Previously, I was a guest researcher on the humanoid TORO team at the German Aerospace Center (DLR) and a visiting student at ETH Zürich. I received my B.S. in Aerospace Engineering from Seoul National University, advised by Prof. Ayoung Kim.

My research bridges model-based knowledge with data-driven learning to overcome data scarcity for robot autonomy.

Email: hojune [AT] stanford.edu

Publication

LEGS: Fine-Tuning Teleop-Free VLAs for Humanoid Loco-manipulation in an Embodied Gaussian Splatting World Hojune Kim, Timothy Chen, Jiankai Sun, Lars W. Osterberg, Qianzhong Chen, Ke Wang, and Mac Schwager arXiv 2026 [Abstract] [arXiv] [Website]
Training vision-language-action (VLA) policies for humanoid loco-manipulation is constrained by the high cost and complexity of collecting human teleoperation demonstrations. VLA policies fine-tuned in simulators have, until now, failed to transfer effectively in humanoid loco-manipulation tasks. We present LEGS (Loco-manipulation via Embodied Gaussian Splatting), a hybrid simulator that composites a mesh foreground (robot, objects, props) over a photorealistic 3D Gaussian Splatting (3DGS) background reconstructed from a handheld scene capture. LEGS uses a procedural motion-primitive generator to synthesize labeled demonstrations at scale without human teleoperation, and a deterministic two-stage color calibration to align the rendered 3DGS image to the robot’s deployment camera. On a Unitree G1 humanoid robot, across three pick-and-place tasks of increasing whole-body difficulty and three VLA backbones, a policy trained purely on LEGS data matches or exceeds one trained on human teleoperation demos on every experiment. It also outperforms a mesh-only simulation baseline that ablates the effect of the 3DGS background, showing that photorealistic rendering is a key enabler for synthetic data transfer. Humanoid motion is recorded independently of scene appearance in LEGS, allowing the same auto-generated demonstrations to be re-rendered under new backgrounds and object meshes–covering a new scene at more than 15x lower cost than teleoperation–to augment training data for robustness to scene variations.

2D Ego-Motion with Yaw Estimation using Only mmWave Radars via Two-Way weighted ICP Hojune Kim, Hyesu Jang, and Ayoung Kim ICRA Workshop Radar in Robotics 2024 [Abstract] [arXiv] [Poster]
The interest in single-chip mmWave Radar is driven by their compact form factor, cost-effectiveness, and robustness under harsh environmental conditions. Despite its promising attributes, the principal limitation of mmWave radar lies in its capacity for autonomous yaw rate estimation. Conventional solutions have often resorted to integrating inertial measurement unit (IMU) or deploying multiple radar units to circumvent this shortcoming. This paper introduces an innovative methodology for two-dimensional ego-motion estimation, focusing on yaw rate deduction, utilizing solely mmWave radar sensors. By applying a weighted Iterated Closest Point (ICP) algorithm to register processed points derived from heatmap data, our method facilitates 2D ego-motion estimation devoid of prior information. Through experimental validation, we verified the effectiveness and promise of our technique for ego-motion estimation using exclusively radar data.

Education

Stanford University
M.S. in Aeronautics and Astronautics GPA: 4.00/4.00

Sept. 2025 - Present

Seoul National University (SNU)
B.S. in Aerospace Engineering, summa cum laude GPA: 4.00/4.00(Major), 3.91/4.00(Overall)

Mar. 2019 - Jun. 2025

ETH Zürich
Visiting Student in Mechanical Engineering

Feb. 2024 - Aug. 2024

Gyeonggi Science High School for the Gifted
1-year early admission

Mar. 2016 - Feb. 2019

Projects

Knowledge-enabled Adaptive Locomotion Using Semantic SLAM with Active Learning German Aerospace Center (DLR) Guest Student Researcher Recursive Estimation, Active Learning, Semantic SLAM

Sept. 2024 - Feb. 2025

Crowd Navigation with LiDAR via Reinforcement Learning Robotic Systems Lab, ETH Zürich [Video] [PDF] Teacher-Student RL, LiDAR Encoding

Mar. 2024 - Jun. 2024

Distributed Optimization with Gaussian Belief Propagation for Continuous-time SLAM Vision for Robotics Lab. ETH Zürich Undergraduate Researcher [PDF] [Slides] [Code] Distributed Optimization, Continuous Time SLAM

Feb. 2024 - Aug. 2024

Display Virtual Cubes on Your Desk! Computer Vision course project, SNU [Video] [PDF] [Code] 3D Vision, Optical Flow

Mar. 2023 - Jun. 2023

Autonomous Flight Landing in Quadrotor system Bulnabi, autonomous flight drone club [Code] Motion Planning, PX4-Autopilot

Jul. 2023 - Aug. 2023

Fault Tolerant Control of Quadcopter Decision Making graduate course project, SNU [PDF] Sliding Mode Control, Backstepping Control, Feedback Linearization

May. 2023 - Jun. 2023

Awards

1st Place, IEEE Humanoids 2024 Adult-sized Biped Walk Competition German Aerospace Center (DLR) Humanoid TORO Team [Video] [Website] Torque Control, Motion Planning, Divergent Component of Motion

Nov. 2024

Minister of National Defense Award, Minister of Defense Startup Competition Addit, Start-up Co-Founder [Video] [News] [Extension] Web Programming, UI/UX Design, Enterprise

Feb. 2022

Gold Prize, International Student Car Competition Autonomous Driving Sector SNU ZERO, autonomous driving car club Sensor Technician [Video] EKF-SLAM, LiDAR-Camera Fusion, Clustering

Oct. 2021

Scholarship

Chungsoo Overseas Scholarship Chungsoo Scholarship Foundation

2025 - 2027

Korea-Germany Junior Research Fellowship Support Max Planck POSTECH/KOREA

Sept. 2024

Global Leadership Program Scholarship Seoul National University

Feb. 2024

Kwanjeong Undergraduate Scholarship Kwanjeong Educational Foundation

Mar. 2021

Undergraduate Research Internship Scholarship Seoul National University

Mar. 2021

Patent

Parking Location Tracking System [KR102291377B1] Hojune Kim, Taekin Kim, Jinhwan Na, Jaeyoung Lee, Seunghwan Jeong

2021

Document editing system supporting online document creation with search function [KR100143446] Hojune Kim, Jinsu Choi

2022

Fun Facts

CH-47D Helicopter Flight Attendant Republic of Korea Army [News] Wild Fire Extinguishment, Helicopter Maintenance, Tactical missions

2021 - 2023

Street Dancer Hoofers in SNU(H.I.S.) Locking, Popping Dance

2019 - 2021

Volunteer

Organized AI Camp and Autonomous Race Car Competition for 200+ Nationwide Students AI Tech Play Co-organizer, Hardware Team Leader [Website] [Youtube]

Feb. 2021 - Aug. 2021