Yunshen Wang

Hi, I’m Yunshen Wang. I am currently a Master’s student at Beijing University of Posts and Telecommunications (BUPT), advised by Prof. Honggang Zhang. I am also a research intern at General Vision Lab in BIGAI (Beijing Institute for General Artificial Intelligence) as a research intern advised by Dr. Siyuan Huang. I received my Bachelor’s degree from BUPT.

Previously, I was a research intern at MARS Lab, where I worked with Prof. Hang Zhao. I have also held internship positions at XREAL, NIO, and Galaxea AI.

My research interests lie at the intersection of Generative Models and Robotics. Specifically, I focus on leveraging generative models for perception, decision-making, planning, and control in robotic systems.

news

Jan 28, 2025	Our paper “Diffusion-Based Generative Models for 3D Occupancy Prediction in Autonomous Driving” has been accepted at ICRA 2025! 🎉 See you in Atlanta! 🌟
Aug 29, 2024	I joined the General Vision Lab at BIGAI (Beijing Institute for General Artificial Intelligence) as a Research Intern, advised by Dr. Siyuan Huang, focusing on Embodied AI and Humanoid Robots.
Jul 01, 2024	I concluded my internship at MARS Lab, thanks to Dr. Yicheng Liu and Prof. Hang Zhao for their supportive and fantastic guidance!

selected publications

Diffusion-Based Generative Models for 3D Occupancy Prediction in Autonomous Driving

Yunshen Wang, Yicheng Liu, Tianyuan Yuan, Yucheng Mao, Yingshi Liang, Xiuyu Yang, Honggang Zhang, and Hang Zhao

2025

Abs arXiv

Accurately predicting 3D occupancy grids from visual inputs is critical for autonomous driving, but current discriminative methods struggle with noisy data, incomplete observations, and the complex structures inherent in 3D scenes. In this work, we reframe 3D occupancy prediction as a generative modeling task using diffusion models, which learn the underlying data distribution and incorporate 3D scene priors. This approach enhances prediction consistency, noise robustness, and better handles the intricacies of 3D spatial structures. Our extensive experiments show that diffusion-based generative models outperform state-of-the-art discriminative approaches, delivering more realistic and accurate occupancy predictions, especially in occluded or low-visibility regions. Moreover, the improved predictions significantly benefit downstream planning tasks, highlighting the practical advantages of our method for real-world autonomous driving applications.
Humanoid Diffusion Controller

Yunshen Wang, Shaohang Zhu, Jingze Zhang, Jiaxin Li, Yixuan Li, Tengyu Liu, and Siyuan Huang

2025

Abs Video Website

We introduce the Humanoid Diffusion Controller, the first diffusion-based generative controller for real-time whole-body control of humanoid robots. Unlike conventional online reinforcement learning (RL) approaches, HDC learns from large-scale offline data and leverages a Diffusion Transformer to generate temporally coherent action sequences. This design provides high expressiveness, scalability, and temporal smoothness. To support training at scale, we propose an effective data collection pipeline and training recipe that avoids costly online rollouts while enabling robust deployment in both simulated and real-world environments. Extensive experiments demonstrate that HDC outperforms state-of-the-art online RL methods in motion tracking accuracy, behavioral quality, and generalization to unseen motions. These findings underscore the feasibility and potential of large-scale generative modeling as a scalable and effective paradigm for generalizable and high-quality humanoid robot control.