Ke Fan

Ke Fan 樊珂

Humanoid Robot Intelligence · Robotics Foundation Models · Embodied AI

I am a fourth-year Ph.D. student at Shanghai Jiao Tong University, advised by Prof. Lizhuang Ma. My research focuses on building foundation-model-driven humanoid robot intelligence across planning, motion generation, whole-body control, and closed-loop feedback.

My current work studies System2--System1--System0 closed-loop learning for general humanoid loco-manipulation, transferring ego-view human operation knowledge into robot-executable actions, and generalizing whole-body control across skills, environments, and manipulation tasks.

Email Google Scholar

Research Framework

Planning--generation--execution--feedback for humanoid robots.

Foundation-model cognition Motion intelligence Embodied execution
System2

Task Planning & Agent Harness

LLM/VLM reasoning, task decomposition, tool use, execution monitoring.

System1

Large Motion / VLA Models

Large motion models, VLA, World Action Model, pretraining and post-training from large-scale human data.

System0

Tracking & Whole-Body Control

Stable control, loco-manipulation, vision-guided feedback, physical constraints.

Human Knowledge Sources

Ego-view operation data · embodiment-free human motion · human video

Robot-Executable Intelligence

General humanoid loco-manipulation across skills, scenes, and manipulation tasks

News

Selected Publications

Grouped by the System2--System1--System0 robotics stack.

Open The Motion Door project thumbnail

Open the Motion Door: Atomic Motion Decomposition and Recomposition for Open-Vocabulary Motion Generation

Ke Fan, Jiangning Zhang, Ran Yi, Jingyu Gong, Yabiao Wang, Yating Wang, Xin Tan, Chengjie Wang, Lizhuang Ma

CVPR 2026 · System1

Go to Zero project thumbnail

Go to Zero: Towards Zero-shot Motion Generation with Million-scale Data

Ke Fan, Shunlin Lu, Minyue Dai, Runyi Yu, Lixing Xiao, Zhiyang Dou, Junting Dong, Zhuangli Ma, Jingbo Wang

ICCV 2025, Highlight · System1

[Project] [Paper]

MotionStreamer project thumbnail

MotionStreamer: Streaming Motion Generation via Diffusion-based Autoregressive Model in Causal Latent Space

Lixing Xiao, Shunlin Lu, Huaijin Pi, Ke Fan, Liang Pan, Yueer Zhou, Ziyong Feng, Xiaowei Zhou, Sida Peng, Jingbo Wang

ICCV 2025 · System1

[Project] [Paper]

FreeMotion project thumbnail

FreeMotion: A Unified Framework for Number-free Text-to-Motion Synthesis

Ke Fan, Junshu Tang, Weijian Cao, Ran Yi, Moran Li, Jingyu Gong, Jiangning Zhang, Yabiao Wang, Chengjie Wang, Lizhuang Ma

ECCV 2024 · System1

[Project] [Paper]

DIP project thumbnail

Diffusion Implicit Policy for Unpaired Scene-aware Motion Synthesis

Jingyu Gong, Chong Zhang, Fengqi Liu, Ke Fan, Qianyu Zhou, Xin Tan, Zhizhong Zhang, Yuan Xie

AAAI 2025 · System1

[Project] [Paper]

TIMotion project thumbnail

TIMotion: Temporal and Interactive Framework for Efficient Human-Human Motion Generation

Yabiao Wang, Shuo Wang, Jiangning Zhang, Ke Fan, Jiafu Wu, Zhucun Xue, Yong Liu

CVPR 2025 · System1

[Project] [Paper]

Unitracker project thumbnail

UniTracker: Learning Universal Whole-Body Motion Tracker for Humanoid Robots

Kangning Yin, Weishuai Zeng, Ke Fan, Minyue Dai, Zirui Wang, Qiang Zhang, Zheng Tian, Jingbo Wang, Jiangmiao Pang, Weinan Zhang

RA-L 2026 · System0

[Project] [Paper]

EGG project thumbnail

EGG: An Expert-Guided Agent Framework for Kernel Generation

Yaochen Han*, Ke Fan*, Hongxu Jiang, Wanqi Xu, Weiyu Xie, Runhua Zhang, Chenhui Zhu, Yixiang Zhang

ICML 2026

[Paper]

Experience

Galbot

Research Intern · May 2026 - Present

Vision-guided System1-to-System0 closed-loop control for humanoid manipulation.

Light Robotics

Research Intern · Jan 2026 - May 2026

Humanoid System1 pretraining and System1-System0 closed-loop co-training.

Shanghai AI Laboratory

Research Intern · Nov 2024 - Dec 2025

Large motion pretraining, humanoid loco-manipulation, and streaming motion generation.

Tencent Youtu Lab

Research Intern · Jan 2022 - Sep 2024

Multi-person and open-vocabulary human motion generation.