About

I am a Master's student in Electrical and Computer Engineering (Robotics track) at the University of Michigan, where I work with Professor Ram Vasudevan in the ROAHM Lab on physically grounded 3D scene reconstruction.

Before that, I earned dual Bachelor's degrees from National Taiwan University, double majoring in Mechanical Engineering and Physics. My undergraduate research experience includes intelligent elderly-care robotics, adviced by Professor Li-Chen Fu at Advanced Control Laboratory, and computational astrophysics with Dr. Hsien Shang at Academia Sinica, Institute of Astonomy and Astrophysics (ASIAA).

I am broadly interested in robust robotic perception and planning, with a focus on developing algorithms that enable robots to achieve human-level perception and understanding of their environments. My current research explores how to integrate robot vision and robot-environment interactions to autonomously generate physically accurate 3D scene representations. In the long term, I aim to enhance robot autonomy and robustness leveraging 3D world models, enabling more robust planning and operation in real-world environments.

News

Nov. 2025	Honored to receive the Rackham International Students Fellowship for the 2025-2026 academic year! See the Department announcement here.
May 2025	Presented my work at ICRA 2025 in Atlanta, GA.

Publications, Presentations, & Workshop Papers

SurfSLAM: Sim-to-Real Underwater Stereo Reconstruction for Real-Time SLAM [Website] [Paper] arXiv preprint arXiv:2601.10814 (2026) Onur Bagoren*, Seth Isaacson*, Sacchin Sundar, Yung-Ching Sun, Anja Sheppard, Haoyu Ma, Abrar Shariff, Ram Vasudevan, Katherine A. Skinner

Key Capabilities of Autonomous Mobile Platforms for Maintenance and Monitoring in Manufacturing Environments [Poster] [Paper] [My Final Report] Late Breaking Results & Workshop on The Future of Intelligent Manufacturing, ICRA 2025 Yung-Ching Sun*, Samantha Staudinger*, Hanna Chapin, Alyssa Carter, Kira Barton, Dawn Tilbury

A Hybrid Acceleration for Self-Gravity Calculation in Infinitesimally Thin Disks with Adaptive Mesh Refinement The Astrophysical Journal Supplement Series (Under review) Chien-Chang Yen, Yung-Ching Sun, Yao-Huan Tseng, Hsien Shang

Selected Projects

Physically Grounded Digital Twin Generation [ Code and Paper Coming Soon... ] Designed an end-to-end framework for physically accurate digital twin reconstruction through a combination of MLLMs reasoning and robot-environment interactive refinement. Simulated the system with NVIDIA Isaac Sim/Lab, deployed in the real-world using TidyBot, finetuned VLA policies and trained RL policies to validate the usage of the reconstructed digital twin.

Autonomous Driving Scene Reconstruction via 3D Gaussian Splatting [ Code ] [ Poster ] Implemented three state-of-the-art 3D Gaussian Splatting-based reconstruction frameworks, Street Gaussians, OmniRe, GaussianSTORM on autonomous driving datasets including Waymo and nuScenes. Performed comprehensive quantitative (PSNR, SSIM, LPIPS) and qualitative evaluations to assess rendering fidelity, geometric stability, and robustness to dynamic objects in complex driving scenarios. Provided guidelines for using these methods, analyzed their strengths and limitations, and outlined potential directions for future research.

Rapid Motor Adaptation (RMA) for RL-Based Quadruped Locomotion Built a two-phase Rapid Motor Adaptation (RMA) pipeline for Unitree Go1 quadruped locomotion in MuJoCo/Brax with JAX-based simulation by extending RLS-RL PPO to consume privileged observations and learn an environment latent, enabling adaptation beyond naive domain randomization. Trained a teacher-student adaptation module via DAgger-style distillation, promoting sim-to-real transfer capabilities, enabling robust locomotion on unseen rough terrains without privileged information at deployment.

Real-Time Visual SLAM for Dynamic Environments [ Code ] [ Report ] [ Poster ] We present a real-time visual SLAM system designed for dynamic environments, leveraging hybrid segmentation with FastSAM and YOLO11n-seg, and optical flow analysis to detect and mask open-vocabulary dynamic feature points, enabling robust and efficient tracking in dynamic scenes. Achieved a maximum of 97.34% reduction in absolute trajectory error compared to the ORB-SLAM3 baseline while maintaining real-time performance, evaluated on TUM and Bonn datasets.

SLAM and Autonomous Exploration for a Differential-Drive Mobile Robot [ Code ] [ Report ] Designed and implemented a complete autonomous system for MBot (a differential-drive wheeled robot). Developed a wheel speed PID controller and a Pure Pursuit motion controller, achieving a final positional error below 3 cm and a heading error under 15° after completing two laps of a 7.308 m closed-loop trajectory. Built a SLAM system using occupancy grid mapping and Monte Carlo Localization, fusing 2D LiDAR, wheel odometry, and IMU data to achieve a 0.0369 m RMS pose error in simulation. Integrated SLAM with A* path planning and frontier-based exploration algorithms, enabling fully autonomous mapping, localization, and navigation in unknown environments.

Vision-Based Manipulation with an RGB-D Sensor and 5-DOF Robotic Arm [ Code ] [ Report ] Developed a vision-guided robotic manipulation system using an RGB-D sensor and a 5-DOF robotic arm, integrating OpenCV-based block detection with kinematics and control for autonomous manipulation. Designed motion-planning strategies enabling block sorting, stacking, and alignment based on shape, size, and color.

ANA*: Anytime Nonparametric A* Algorithm [ Code ] [ Report ] Implemented and evaluated A* and ANA* algorithms for PR2 robot navigation, developing custom heuristic functions and 3D PyBullet environments to benchmark performance. Demonstrated ANA*'s ability to rapidly produce suboptimal paths and iteratively reach optimal solutions, outperforming A* across varied scenarios.

Experience

	Graduate Research Assistant ROAHM (Robotics and Optimization for the Analysis of Human Motion) Lab, University of Michigan, Feb. 2025 - Present
	Graduate Research Assistant Barton Research Group, University of Michigan, Jan. 2025 - May 2025
	Undergraduate Researcher Advanced Control Lab - Intelligent Robotics Group, National Taiwan University, Sep. 2022 - Aug. 2024
	Undergraduate Research Intern Academia Sinica, Institute of Astronomy and Astrophysics, Jul. 2022 - Jul. 2024
	Rocket Avionics Hardware Engineer Summer Intern Advanced Rocket Research Center, Taiwan, Jul. 2021 - Sep. 2021

Education

	University of Michigan - Ann Arbor M.S. in Electrical and Computer Engineering (Robotics), Aug. 2024 - Present
	National Taiwan University B.S.E in Mechanical Engineering & B.S. in Physics, Sep. 2019 - Jun. 2023

Honors & Awards

Rackham International Student Fellowship Rackham Graduate School, University of Michigan, Nov. 2025

Academia Sinica Undergraduate Research Scholarship Academia Sinica, Institute of Astronomy and Astrophysics, Sep. 2022