Yidan Lu (鲁一丹)

personal_imgs/gyl2.png

Hong Kong SAR, China. 999077

Email: ydlu@connect.hku.hk

I am a PhD student at The University of Hong Kong, affiliated with the ArcLab under the supervision of Professor Peng Lu. My research lies at the intersection of reinforcement learning and legged robotics, with a focus on enabling humanoid robots to learn from real-world interactions and acquire human-like capabilities.

I earned my Bachelor’s degree from Northwestern Polytechnical University in 2023, where I conducted research in audio speech recognition (ASR) under the guidance of Professor Lei Xie at the ASLP Lab. My work focused on enhancing the accuracy and efficiency of speech recognition systems.

I am passionate about advancing robotics to bridge the gap between technology and human society, aiming to automate repetitive tasks and unlock new possibilities for human potential.

Email / Google Scholar / Github / Linkedin

news

Apr 24, 2025 Our recent paper “FR-Net: Learning Robust Quadrupedal Fall Recovery on Challenging Terrains through Mass-Contact Prediction” has been accepted by RAL! :sparkles: :smile:
Dec 30, 2024 Honored to Meet Richard S. Sutton at CUHK!
Sep 15, 2024 Honored to have Prof. Toshio Fukuda visiting our LAB!

Publications

sym

Contrastive Representation Learning for Robust Sim-to-Real Transfer of Adaptive Humanoid Locomotion
Yidan Lu1,2*, Rurui Yang2*, Qiran Kou2*, Mengting Chen2†, Peter Cui2, Yinzhao Dong1, Peng Lu,

Arxiv 2025

A learning based method for humanoid robot walking over challenging terrains.
sym

FR-Net: Learning Robust Quadrupedal Fall Recovery on Challenging Terrains through Mass-Contact Prediction
Yidan Lu*, Yinzhao Dong*, Jiahui Zhang, Ji Ma, Peng Lu†,

IEEE Robotics and Automation Letters (RAL) 2025
IROS 2025 Presentation

A learning based method for quadrupedal robot robust recovery on challenging terrains.

Projects

sym

Vision-guided Adaptive Locomotion for Quadrupedal Robots Using Reinforcement Learning
Team members: Yidan Lu, Yinzhao Dong, Ji Ma, Jiahui Zhang
Supervised by Peng Lu

A robust learning-based control framework that enables adaptive quadrupedal locomotion on challenging terrains through depth camera perception, 2025.