Yu Guo

Yu Guo 郭语

MS Student @ UCLA ECE

Embodied AI Probabilistic Modeling Signal Processing World Models

About Me

I am currently an ECE Master's student at UCLA ('26). My work focuses on bridging the gap between intelligence in digital world and physical reality.

My research focuses on Physical AI — building unified models that integrate perception and action for real-world interaction. I develop point-cloud-based world models and video-action models to learn structured 3D dynamics for scalable robotic control.

Open to Opportunities: I am actively seeking industry R&D internships or Gap Year research positions (Starting Summer 2026). I am passionate about scaling robot learning and building robust foundation models for physical agents. Let's chat

University of California, Los Angeles

M.S. in Electrical and Computer Engineering (2024-2026)

UESTC

B.Eng. in Communication Engineering (2020-2024)

Research Focus

Current Frontiers

Point-Cloud World Models

Developing generative world models operating directly on 3D point clouds to capture spatial dynamics and occlusion-robust latent states.

Video-Action Foundation

Fine-tuning visual foundation models for direct action token prediction, enabling agents to learn complex manipulation from observation.

News

JAN 2026

Interning at UCLA AIVC Lab.

JAN 2026

Paper accepted at WCNC 2026.

NOV 2025

Paper accepted at ISNCC 2025.

Research Background

Extensive experience in probabilistic modeling and advanced signal processing applied to satellite networks and neural systems.

2024 - 2025

LEO Satellite Routing

Proposed a probabilistic feature fusion and non-convex optimization framework for robust routing. Focused on modeling link reliability under high dynamic topologies. Research intern at National Key Laboratory of Wireless Communication. Probabilistic Modeling

VIEW PAPER (ISNCC 2025)
2025

Neural Signal Decoding

Developed a signal processing pipeline for Brain-to-Text translation using dual-mask augmentation and consistency-regularized CTC. Project at UCLA. Signal Processing

VIEW CODE
2025

Semantic Communication

Decomposed instruction-oriented text for high-fidelity communication. Research intern at Tsinghua University.