Wenhao Yu

I am currently pursuing a PhD at University of Science and Technology of China (USTC) advised by Prof. Yanyong Zhang. My M.S. is also from USTC and I was advised by Prof. Jianmin Ji. Before that, I received my Bachelor's degree in Computer Science from Qingdao University (QDU) in 2021.

I'm broadly interested in AI for robotics and more focused on learning-based robot decision-making and planning: Robot Navigation, Autonomous Driving, and Embodied AI.

News

• [2025.4] Our paper is accepted by RSS 2025.

• [2024.11] I am looking for a research intern on embodied AI and ML. If you have any related needs, please contact me. Thank you very much!

• [2024.6] Our paper is accepted by IROS 2024. Looking forward to seeing you in Abu Dhabi, UAE.

• [2024.1] Our paper is accepted by ICRA 2024. Looking forward to seeing you in Yokohama, Japan.

Publications

	STDArm: Transfer Visuomotor Policy From Static Data Training to Dynamic Robot Manipulation Yifan Duan, Heng Li, Yilong Wu, Wenhao Yu, Xinran Zhang, Yedong Shen, Jianmin Ji, Yanyong Zhang, RSS*, 2025 project page / code / arXiv This paper presents STDArm, a system that transfers static-trained visuomotor policies to dynamic robots, achieving precise and stable manipulation through real-time action correction without retraining.
	MHRC: Closed-loop Decentralized Multi-Heterogeneous Robot Collaboration with Large Language Models Wenhao Yu, Jie Peng, Yueliang Ying, Sai Li, Jianmin Ji, Yanyong Zhang, arXiv*, 2409.16030 project page / code / arXiv LLMs are used to realize the collaboration of multiple heterogeneous robots (mobile robot, manipulation robot, and mobile manipulation robot), including three tasks: make sandwich, sort solids, and pack objects.
	LDP: A Local Diffusion Planner for Efficient Robot Navigation and Collision Avoidance Wenhao Yu, Jie Peng, Huanyu Yang, Junrui Zhang, Yifan Duan, Jianmin Ji, Yanyong Zhang, IROS*, 2024 project page / code / arXiv Model the multi-modal expert policy distribution with multiple scenarios and preferences by diffusion model for robot navigation and collision avoidance.
	PathRL: An End-to-End Path Generation Method for Collision Avoidance via Deep Reinforcement Learning Wenhao Yu, Jie Peng, Quecheng Qiu, Hanyu Wang, Lu Zhang, Jianmin Ji, ICRA*, 2024 project page / code / arXiv a novel end-to-end DRL-based method, PathRL, that directly outputs navigation paths without relying on the supervised learning paradigm and iscompetent for a variety of complex scenarios.

Projects

Build different types of real navigation systems
Wenhao Yu

Build and deploy robot decision-making and planning systems on different types of mobile chassis, including differential chassis, Ackerman chassis, logistics vehicles, passenger cars, etc. For more details please refer to my resume.

The website template is adapted from Jon Barron's website.