Yicheng Liu | Tsinghua University

About Me

I am a Ph.D. candidate at the Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University, advised by Prof. Hang Zhao. Before that, I received my Master's degree from the Chinese University of Hong Kong and my Bachelor's degree from Xidian University, where I was advised by Prof. Bo Chen.

My goal is to build machines that perceive, reason about, and act in the physical world. I am currently working on embodied foundation models — large Vision-Language-Action (VLA) models pretrained across embodiments, scenes, and tasks, so that a single model can follow open-ended language instructions and manipulate unseen objects out of the box. My recent work centers on neural action tokenization and efficient autoregressive VLA architectures (FASTer, ActionCodec), which underpin the Galaxea G0/G0.5 model family. Earlier, I worked on end-to-end autonomous driving and online HD map learning (VectorMapNet, Neural Map Prior, StreamMapNet, DriveVLM).

Research Interests

Embodied Foundation Models: VLA architectures, action tokenization, large-scale pretraining, post-training
Generative Models for Decision Making: diffusion & flow policies, world-action models
Autonomous Driving: end-to-end learning, online mapping, vision-language models

News

[Aug. 2026] The Galaxea G0.5 technical report is now on arXiv.
[May 2026] We released Galaxea G0.5, a zero-shot embodied foundation model. A single transformer decoder emits both reasoning and action tokens under one objective, keeping the VLM the decision-maker rather than just a context encoder.
[Mar. 2026] We released Fast-WAM, a world-action model that removes explicit future imagination at test time.
[Feb. 2026] We released ActionCodec, a systematic study of what makes a good action tokenizer for VLA models.
[Jan. 2026] DepthVLA and the Galaxea Open-World Dataset are accepted by ICRA 2026.
[Dec. 2025] We released FASTer, a neural action tokenizer for efficient autoregressive VLA modeling.
[Sep. 2025] Our paper Cocos is accepted by NeurIPS 2025.
[Sep. 2025] We released the Galaxea Open-World Dataset and G0 VLA on GitHub and Hugging Face.
[Sep. 2024] Our paper DriveVLM is accepted by CoRL 2024.

Publications (* indicates equal contribution)

Galaxea G0.5: One Autoregressive Stream for Robot Reasoning and Action

Galaxea AI Research

arXiv:2608.11739, 2026.

arXiv PDF Project Page
Fast-WAM: Do World Action Models Need Test-time Future Imagination?

Tianyuan Yuan, Zibin Dong, Yicheng Liu, Hang Zhao

arXiv preprint, 2026.

PDF Code Project Page
ActionCodec: What Makes for Good Action Tokenizers

Zibin Dong, Yicheng Liu, Shiduo Zhang, Baijun Ye, Yifu Yuan, Fei Ni, Jingjing Gong, Xipeng Qiu, Hang Zhao, Yinchuan Li, Jianye Hao

arXiv preprint, 2026.

PDF
FASTer: Toward Efficient Autoregressive Vision-Language-Action Modeling via Neural Action Tokenization

Yicheng Liu, Shiduo Zhang, Zibin Dong, Baijun Ye, Tianyuan Yuan, Xiaopeng Yu, Linqi Yin, Chenhao Lu, Junhao Shi, Luca Jiang-Tao Yu, Liangtao Zheng, Tao Jiang, Jingjing Gong, Xipeng Qiu, Hang Zhao

arXiv preprint, 2025.

PDF
ICRA

DepthVLA: Enhancing Vision-Language-Action Models with Depth-Aware Spatial Reasoning

Tianyuan Yuan, Yicheng Liu, Chenhao Lu, Zhuoguang Chen, Tao Jiang, Hang Zhao

IEEE International Conference on Robotics and Automation (ICRA), 2026.

PDF
ICRA

Galaxea Open-World Dataset and G0 Dual-System VLA Model

Tao Jiang*, Tianyuan Yuan*, Yicheng Liu*, Chenhao Lu, Jianning Cui, Xiao Liu, Shuiqi Cheng, Jiyang Gao, Huazhe Xu, Hang Zhao

IEEE International Conference on Robotics and Automation (ICRA), 2026.

PDF Code Project Page
NeurIPS

Conditioning Matters: Training Diffusion Policies is Faster Than You Think

Zibin Dong, Yicheng Liu, Yinchuan Li, Hang Zhao, Jianye Hao

Conference on Neural Information Processing Systems (NeurIPS), 2025.

PDF
ICRA

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

Yunshen Wang*, Yicheng Liu*, Tianyuan Yuan*, Yingshi Liang, Xiuyu Yang, Honggang Zhang, Hang Zhao

IEEE International Conference on Robotics and Automation (ICRA), 2025.

PDF
CoRL

DriveVLM: The Convergence of Autonomous Driving and Large Vision-Language Models

Xiaoyu Tian, Junru Gu, Bailin Li, Yicheng Liu, Yang Wang, Zhiyong Zhao, Kun Zhan, Peng Jia, Xianpeng Lang, Hang Zhao

Conference on Robot Learning (CoRL), 2024.

PDF Project Page
ECCV

PreSight: Enhancing Autonomous Vehicle Perception with City-Scale NeRF Priors

Tianyuan Yuan, Yucheng Mao, Jiawei Yang, Yicheng Liu, Yue Wang, Hang Zhao

European Conference on Computer Vision (ECCV), 2024.

PDF Code
WACV

StreamMapNet: Streaming Mapping Network for Vectorized Online HD Map Construction

Tianyuan Yuan, Yicheng Liu, Yue Wang, Yilun Wang, Hang Zhao

Winter Conference on Applications of Computer Vision (WACV), 2024.

PDF Code
CoRL

Cross-Dataset Sensor Alignment: Making Visual 3D Object Detector Generalizable

Liangtao Zheng, Yicheng Liu, Yue Wang, Hang Zhao

Conference on Robot Learning (CoRL), 2023.

PDF
CVPR

Neural Map Prior for Autonomous Driving

Xuan Xiong, Yicheng Liu, Tianyuan Yuan, Yue Wang, Yilun Wang, Hang Zhao

Conference on Computer Vision and Pattern Recognition (CVPR), 2023.

PDF Code
ICML

VectorMapNet: End-to-End Vectorized HD Map Learning

Yicheng Liu, Tianyuan Yuan, Yue Wang, Yilun Wang, Hang Zhao

International Conference on Machine Learning (ICML), 2023.

PDF Code Project Page
CVPR

Multimodal Motion Prediction with Stacked Transformers

Yicheng Liu, Jinghuai Zhang, Liangji Fang, Qinhong Jiang, Bolei Zhou

Conference on Computer Vision and Pattern Recognition (CVPR), 2021.

PDF Code Project Page

Earlier work (machine learning & signal processing)

Balanced Gradient Penalty Improves Deep Long-Tailed Learning
Dong Wang, Yicheng Liu, Liangji Fang, Fanhua Shang, Yuanyuan Liu, Hongying Liu. ACM International Conference on Multimedia (ACM MM), 2022.
Infinite Switching Dynamic Probabilistic Network with Bayesian Nonparametric Learning
Wenchao Chen, Bo Chen, Yicheng Liu, Chaojie Wang, Xiaojun Peng, Hongwei Liu, Mingyuan Zhou. IEEE Transactions on Signal Processing, 2022.
PSNet: Perspective-Sensitive Convolutional Network for Object Detection
Xiuqi Zhang, Yicheng Liu, Chunlei Huo, Nuo Xu, Lingfeng Wang, Chunhong Pan. Neurocomputing, 2022.
Bidirectional Recurrent Gamma Belief Network for HRRP Target Recognition
Wenchao Chen, Bo Chen, Yicheng Liu, Xiaojun Peng, Hao Fan, Fan Yu, Hongwei Liu. Signal Processing, 2021.
Bidirectional Convolutional Poisson Gamma Dynamical Systems
Wenchao Chen, Chaojie Wang, Bo Chen, Yicheng Liu, Hao Zhang, Mingyuan Zhou. Conference on Neural Information Processing Systems (NeurIPS), 2020.
Switching Poisson Gamma Dynamical Systems
Wenchao Chen, Bo Chen, Yicheng Liu, Qianru Zhao, Mingyuan Zhou. International Joint Conference on Artificial Intelligence (IJCAI), 2020.
SignADAM++: Learning Confidences for Deep Neural Networks
Dong Wang, Yicheng Liu, Wenwo Tang, Fanhua Shang, Hongying Liu, Qigong Sun, Licheng Jiao. IEEE International Conference on Data Mining Workshops (ICDMW), 2019.

Academic Service

Reviewer: CVPR, IEEE Robotics and Automation Letters (RA-L), IEEE Transactions on Signal Processing (TSP)
Teaching Assistant: Introduction to Multimodal Learning, Tsinghua University (Spring 2024)