Jiajun Liu

Incoming Ph.D. Student, Institute for Interdisciplinary Information Sciences, Tsinghua University

Undergraduate, Gaoling School of Artificial Intelligence, Renmin University of China

About

Hi there 👋🏻, I am Jiajun Liu (刘嘉俊), a senior undergraduate student at the Gaoling School of Artificial Intelligence, Renmin University. I will join the Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University as a Ph.D. student in Fall 2027, advised by Prof. Mengdi Xu with a current focus on robot learning from non-expert data.

Previously, I spent a rewarding summer at MLL Lab, Northwestern University, working with Kangrui Wang, Zihan Wang and Prof. Manling Li. Before that, I was fortunate to work with Prof. Chongxuan Li and Prof. Jun Zhu on 3D/video world models.

My research interests lie in spatial intelligence, world models and embodied AI. I am particularly interested in building learning systems that enable embodied agents to acquire useful behaviors from diverse data and act reliably in the physical world. I am always open to collaborations, discussions, and new connections.

News

2026.10Grateful to receive the NeurIPS 2026 Financial Assistance Award. See you in Sydney this December! :)
2026.09CATok was accepted to NeurIPS 2026, and DEX-X was accepted to CoRL 2026!
2026.09Awarded the China National Petroleum Corporation Scholarship (CNPC Scholarship, ￥12,000).
2026.06Honored with the SenseTime Scholarship Nomination Award. Congrats to my friend Haoxiang on receiving the SenseTime Scholarship! We are the first students from RUC to receive this recognition.
2026.04Embodied Reasoner was accepted to ACL 2026.
2026.02Bronze Medal at the 50-th ICPC East-Asia Continent Final (EC-Final) at Hangzhou.
2025.12Awarded Most Potential Award at Open Source Promotion Plan (OSPP) 2025.
2025.09Honored with National Scholarship (Top 0.4% Nationwide).

Research Experiences

Oct 2025 - Present

Research Intern, IIIS, Tsinghua University, China

Advisor: Mengdi Xu

Jun 2025 - Sep 2025

Research Intern, MLL-Lab, Northwestern University, USA

Advisor: Manling Li

Jan 2025 - Aug 2025

Research Intern, GSAI-ML Group, Renmin University, China

Advisor: Chongxuan Li, Jun Zhu

Publications & Preprints

* denotes equal contribution; † denotes corresponding author.

RoboCoach: World Models as Active Coaches for Compositional Robot Skills

Jiajun Liu*, Yifan Chen*, Yichao Liu*, Jiayi Zhang, Ruoqu Chen, Shaoxuan Xie, Guocai Yao, Mengdi Xu†, Sen Cui†, Changshui Zhang†

Preprint

arXiv Project Code Hugging Face

What Do VLAs Actually Learn through In-Context Failure Conditioning?

Jiajun Liu, Jieming Li, Zi Zhuang, Hang Yu, Qingli Chen, Liu Cao, Yingxi Lu, Ruoqu Chen, Yuhang Cao, Chenyu Zhang, Yankai Lin, Mengdi Xu†

CVPR 2026 Workshop on 3D-LLM/VLA

Project Code

Geometry-Aware Rotary Position Embedding for Consistent Video World Model

Chendong Xiang*, Jiajun Liu*, Jintao Zhang*, Xiao Yang, Zhengwei Fang, Shizun Wang, Zijun Wang, Yingtian Zou, Hang Su†, Jun Zhu†

ICLR 2026 Workshop on World Models

arXiv Code

HuMiT: Low-Latency Whole-Body Humanoid Teleoperation via Minimal Reference Tracking

Liu Cao, Botian Xu, Ruoqu Chen, Jiajun Liu, Mingzhi Pei, Zihao Wang, Zhongyu Li†, Mengdi Xu†

Preprint

Project

Rethinking Causal Action Tokenization with Conditional Annealing in Flow Matching

Chenyu Zhang*, Yuhang Cao*, Daru Du, Yingxi Lu, Jing Shao, Ruoqu Chen, Jiajun Liu, Liu Cao, Yicheng Liu, Hang Zhao, Mengdi Xu†

Annual Conference on Neural Information Processing Systems (NeurIPS), 2026 · Oral @ NeurIPS 2026 WRL

arXiv Project Code

Learning Visual-Tactile Dexterous Manipulation From Human Videos with Simulated Interaction

Ruoqu Chen, Feixiang Ruan, Liu Cao, Zihao Wang, Botian Xu, Shiqin Tong, Jiajun Liu, Wanli Xing, Kaifeng Zhang, Mengdi Xu†

Conference on Robot Learning (CoRL), 2026 · Spotlight @ RSS 2026 DexManip

Project

Embodied Reasoner: Synergizing Visual Search, Reasoning, and Action for Embodied Interactive Tasks

Wenqi Zhang*, Mengna Wang*, Gangao Liu, Huixin Xu, Yiwei Jiang, Yongliang Shen, Guiyang Hou, Zhe Zheng, Hang Zhang, Xin Li, Jiajun Liu, Weiming Lu, Peng Li†, Yueting Zhuang†

Annual Meeting of the Association for Computational Linguistics (ACL), 2026

arXiv Project Code

Selected Open-Source Projects

Embodied Reasoner (With OSPP's funding)

Main Contributor

Project Code

Embodied-Reasoner (ER.) is a multimodal model designed for deep reasoning and long-horizon interaction. Through OSPP, similar to GSoC, I was selected by the AGIROS Community as the contributor responsible for ER. I took charge of benchmarking ER. on ALFRED and contributed to resolving two key bottlenecks: ambiguity in distinguishing identical object instances and imprecise targeting of large objects, thereby improving spatial accuracy and interaction robustness.

RAGEN & VAGEN: Training Agents by Reinforcing Reasoning

Contributor

Project Code

This twin of projects empower agents with RL to operate effectively in interactive and stochastic environments by handling multi-turn interactions and environmental uncertainty. I contributed to developing more environments and mask functions to compute the loss only for the parts generated by the model, which actually made training more stable.

Selected Awards

Most Potential Award, Open Source Promotion Plan (OSPP) 2025 Dec 2025
Gold Medal, International Collegiate Programming Contest (ICPC) Asia Regional (Wuhan) Nov 2025
Gold Medal, "Xiaomi Cup" China Collegiate Programming Contest (CCPC) Invitational Contest Apr 2025
Silver Medal, International Collegiate Programming Contest (ICPC) East-Asia Continent Final Dec 2024
Silver Medal, 2024 CCF Collegiate Computer Systems & Programming Contest (CCSP) Oct 2024

Selected Scholarships

China National Petroleum Corporation Scholarship Sep 2026
SenseTime Scholarship Nomination Jun 2026
National Scholarship Sep 2025
"Linghang" Intellectual Excellence Dean's Scholarship Dec 2024
Outstanding Student Leader Scholarship Sep 2024

Services & Presentations

Reviewer: ICLR 2027
Invited Video: "My Experience Using AI+ Tools to Create Videos"
Invited by CCF for China National Computer Conference (CNCC) Super Forum
Invited Talk: "Algorithm and Artificial Intelligence"
Invited by Gaoling School of AI & School of Information, Renmin University of China

Interests & Hobbies

Basketball

I enjoy playing basketball, as it keeps me active and sharpens my teamwork skills. It has taught me the value of strategy, coordination, and the importance of consistent practice and persistence. I once won the All-Star championship in a Freshman basketball tournament.

Sailing

Sailing represents my love for adventure and the sea. It's a sport that requires patience, understanding of weather patterns, and the ability to work with natural elements. I once won the runner-up in the National Amateur Sailing Competition.

Music

Music is my creative outlet and a way to unwind. I enjoy a wide range of genres — music helps me stay focused during work and sparks inspiration for my projects. I also love singing, and my favorite artists are Taylor Swift and Ed Sheeran.

Coding Contest

Unlike most algorithm competition medalists, I had no OI experience in high school. Fortunately, I found two like-minded teammates who also started from scratch. Together, we trained and grew. Programming contests like ICPC challenged me intellectually and ignited my passion throughout college.

Video Editing

Video editing allows me to express my creativity through visual storytelling. I enjoy combining footage, music, and especially visual effects to craft compelling content. Feel free to follow me on Bilibili.

Last Updated: Oct, 2026