Hi, I'm Lyuxing!

Undergraduate Researcher

Human-Centered Autonomy Lab · Prof. Katie Driggs-Campbell

Contributed to a marginalized importance sampling framework that combines real-world offline data with imperfect simulators for stable off-environment policy evaluation. The framework achieved state-of-the-art results in Sim2Sim benchmarks and was validated through Sim2Real experiments on a Kinova robotic arm.

Research Assistant

Mitra Research Group · Prof. Sayan Mitra

Developed a web-based debugger for Verse, the group's autonomous-system verification tool, to support the design and analysis of autonomous vehicles. Evaluated Verse on a simulated GEM vehicle in Gazebo and on a physical GEM vehicle.

Undergraduate Researcher

National Center for Supercomputing Applications · Prof. André Schleife

Built volumetric models from raw electron-density data and an interactive holographic rendering system controlled by remote hand gestures. Tuned the system for classroom teaching and outreach demonstrations to make materials science concepts easier to understand. Received NCSA recognition for outstanding undergraduate research.

Lead Electrical Engineer

Illini Air Shuttle

Led electrical engineering work for student-built, small-scale eVTOL prototypes. Organized discussions on design constraints and thrust and weight estimates, and coordinated PCB design, 3D-printed component assembly, electronics soldering, and vertical-takeoff testing. Taught seminars on robot control systems and computer vision using the first working subscale model.

Undergraduate Researcher

Ren Lab · Prof. Hongliang Ren

Built and deployed a semi-autonomous, ROS-controlled data collection system using an Olympus duodenoscope, producing the first in-vivo, in-cavity gastrointestinal dataset with complete ground truth for dense 3D reconstruction. Designed a monocular NeRF framework that separates diffuse and specular radiance to handle near-field lighting, fluids, and exposure changes in endoscopy. This work was published in Handbook of Robotic Surgery.

Research Assistant

Robots Perceiving and Doing Lab (R-PAD) · Prof. David Held

Developed a hierarchical, object-centric point diffusion framework combining dense Gaussian mixture model (GMM) global initialization with disentangled geometry and frame diffusion, achieving state-of-the-art placement precision, multimodal coverage, and generalization across rigid and non-rigid placement tasks. This work was accepted to ICRA 2026.

Developed ParticleSplat, a self-supervised, object-centric 3D representation learning framework that maps multi-view observations into compact latent particles and particle-aligned Gaussian splats. Trained through novel-view synthesis without segmentation supervision, the framework enables 3D object decomposition and controllable scene editing, while improving downstream robotic manipulation policies. This work is available as a preprint.

Electrical Engineering Intern

Helped automate scoring for the MakeX Robotics Competition and led MakeX Starter competition design under the department manager's supervision. Assisted with parameter validation and CAD work for CyberPi microcontrollers, sensors, and other Makeblock electronics.

Research Intern

Developed a Real2Sim framework using 4D Gaussian splatting and feature splatting to transform in-the-wild videos into simulation environments. Generated grounded reward functions to support scalable training of robotic manipulation foundation models beyond traditional human-demonstration pipelines.

Senior Machine Learning EngineerCurrent

Founding Machine Learning Engineer

Designed and built SafeWorld’s first human behavior modeling system from the ground up, enabling customers to simulate robot–human interactions and uncover risks before deployment. Contributed to AI-powered simulation and safety testing systems that help make robots safe around people.