I am an undergraduate studying Data Science and Big Data Technology at Huazhong University of Science and Technology (HUST). My research interests are in robot learning and reinforcement learning, with a focus on whole-body control and humanoid loco-manipulation.我是华中科技大学数据科学与大数据技术专业本科生，研究兴趣是机器人学习与强化学习，关注全身控制和人形机器人移动操作。
[2026/07]🎉 I2E appears at ACL 2026!🎉 I2E 发表于 ACL 2026！
[2025/10]🏆 Our project received the National Grand Prize in the 19th Challenge Cup Open Challenge Competition.🏆 项目获第十九届“挑战杯”揭榜挂帅擂台赛国家级特等奖。
Research研究
My research interests are in Robot Learning and Reinforcement Learning, especially whole-body control and humanoid loco-manipulation. I am interested in learning reusable behavioral representations and skills that support generalization across robots and tasks.我的研究兴趣是机器人学习与强化学习，尤其关注全身控制和人形机器人移动操作。我希望学习可复用的行为表征与技能，支持机器人在不同本体和任务中的泛化。
@misc{li2026mimicall,
title={MimicAll: Structured Embodiment Conditioning for Cross-Humanoid Whole-Body Motion Tracking},
author={Jia Li and Lingyun Xu and Yitang Li and Hao Zhao},
year={2026},
eprint={2610.33508},
archivePrefix={arXiv},
primaryClass={cs.RO},
url={https://arxiv.org/abs/2610.33508}
}
arXiv preprint, 2026arXiv 预印本，2026
Structured embodiment conditioning for a shared motion-tracking policy across 16 humanoid families. I led the method, training infrastructure, benchmark construction, experiments and writing.通过结构化本体条件化，让一个动作跟踪策略适配 16 个人形机器人系列。我负责方法设计、训练系统、基准构建、实验与论文撰写。
A framework for learning reusable humanoid skills without task-specific rewards. I contributed to the unsupervised RL research on forward-backward representations, temporal contrastive learning and whole-body control.研究无需任务特定奖励的可复用人形机器人技能。我参与无监督强化学习框架研究，涉及前向—后向表征、时序对比学习与全身控制。
@misc{wu2026sugar,
title={SUGAR: A Scalable Human-Video-Driven Generalizable Humanoid Loco-Manipulation Learning Framework},
author={Tianshu Wu and Xiangqi Kong and Yue Chen and Qize Yu and Hang Ye and Jia Li and Yizhou Wang and Hao Dong},
year={2026},
eprint={2605.20373},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.20373}
}
NeurIPS 2026NeurIPS 2026
Learning humanoid loco-manipulation from human videos. My work covered human–object interaction priors, physics-based demonstration refinement and the hierarchical controller and evaluation pipeline.从人类视频学习人形机器人的移动操作。我参与人—物交互先验提取、物理约束演示优化，以及分层控制与评测流程。
@inproceedings{yu2026i2e,
title={{I2E}: From Image Pixels to Actionable Interactive Environments for Text-Guided Image Editing},
author={Jinghan Yu and Junhao Xiao and Chenyu Zhu and Jiaming Li and Jia Li and HanMing Deng and Xirui Wang and Guoli Jia and Jianjun Li and Xiang Bai and Bowen Zhou and Zhiyuan Ma},
booktitle={Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
year={2026},
pages={44827--44850},
doi={10.18653/v1/2026.acl-long.2076},
url={https://aclanthology.org/2026.acl-long.2076/}
}
ACL 2026ACL 2026
Turning image editing into interactions with a decomposed environment. I worked on background inpainting, mask refinement, environment-aware semantic integration and benchmark evaluation.将图像编辑转化为与分解后环境的交互。我负责背景修复分支、掩码优化、环境语义融合，并参与基准评测。
* Equal contribution. My name is underlined in the author lists. Hover over a demo thumbnail to play, or use its Play demo button.* 表示共同贡献；作者列表中以粗体下划线标出我的名字。悬停演示缩略图可播放，也可使用“播放演示”按钮。
Mar – Jun 20262026.03 – 2026.06 · Core contributor核心参与者
Built research-skill tagging and retrieval, automated ingestion with retries and idempotent database updates and deployment with systemd and Nginx.参与研究技能标签与检索系统、支持重试与幂等入库的自动化处理流程，以及 systemd、Nginx 服务部署。
LingShu · General-purpose AI assistant灵枢 · 通用 AI 超级助手
Jun – Aug 20252025.06 – 2025.08 · Project contributor项目参与者
Helped build a reusable library of 368 atomic actions across 29 apps, with structured action definitions and tests for cross-app task execution.参与构建覆盖 29 类 App、368 个原子操作的可复用动作库，整理结构化动作描述与测试，支持跨应用任务执行。
Shuzheng e-Station · Government-service AI agent数政 e 站 · 数据协同政务智能体
Jun – Aug 20252025.06 – 2025.08 · Project contributor项目参与者
Contributed domain knowledge integration, structured SFT data and benchmarks, vLLM inference and API/database integration for a government-service agent on the Jiutian platform.依托“九天”大模型平台，参与政务知识接入、结构化 SFT 数据与基准构建、vLLM 推理，以及 API 和数据库集成。
Sep 2024 – Jun 2028 (expected)2024.09 – 2028.06（预计）
Undergraduate, Data Science and Big Data Technology数据科学与大数据技术 · 本科
Overall score: 90.5/100 · Major rank: 1/70综合成绩：90.5/100 · 专业排名：1/70
Selected coursework: Machine Learning (95), Introduction to NLP (95), Algorithm Design and Analysis Practice (97), Database Systems Practice (97) and Computer Systems Lab (99).部分课程：机器学习导论（95）、NLP 导论（95）、算法设计与分析实践（97）、数据库系统原理实践（97）、计算机系统基础实验（99）。
Served as the academic affairs representative of class CS2404 (2024–2025) and joined the School of Computer Science alumni affairs center and Yuhui Workshop in 2025.曾任 CS2404 班学习委员（2024–2025），2025 年起参与计算机学院校友工作中心与“喻慧工坊”。
More than 66 hours of volunteer service, including school outreach, campus events, botanical-garden volunteering, hospital guidance and community activities.累计志愿服务 66 余小时，参与母校宣讲、校园活动、植物园义工、医院导诊与社区服务。
Self-directed study includes CS231n and CS285, alongside hands-on reproduction of robot-learning and reinforcement-learning methods.自主学习 CS231n、CS285 等课程，并通过机器人学习与强化学习方法复现积累实践经验。
Tools I work with:常用工具： Python, C/C++, PyTorch, Isaac Lab / mjlab, Unitree RL Lab, Git, SQL, LaTeX.
