Zenghuang Fu

Institute of Automation

Chinese Academy of Sciences

Beijing, China

I am a master’s student at the Institute of Automation, Chinese Academy of Sciences (CASIA), advised by Prof. Weiliang Meng and Prof. Xiaopeng Zhang.

My research interests are long-horizon credit assignment for agentic reinforcement learning, agentic search, self-evolving agents and AutoResearch. I have a background in agentic memory, agentic RL and long-video understanding with multimodal LLMs.

I have contributed to work published at AAAI 2026, ACL 2026, Information Fusion, Expert Systems with Applications and The Visual Computer. A full list is on my publications page.

I maintain two reading lists: Awesome-Long-Horizon-Credit-Assignment-for-Agentic-RL, on long-horizon credit assignment for agentic RL, and Awesome-Agentic-OPD, on on-policy distillation and self-distillation for multi-turn agents.

As an undergraduate I was a National First Prize winner in the China Undergraduate Mathematical Contest in Modeling (CUMCM), a Meritorious Winner in the Mathematical Contest in Modeling (MCM/ICM), and won a Fujian Provincial First Prize in the national Zhengda Cup market survey and analysis competition — all in 2022. The full list is on my CV.

I am actively looking for PhD positions. If my interests line up with your group’s, I would be glad to hear from you — please feel free to reach out by email.

news

Aug 03, 2026	Two search-agent preprints on arXiv: SESA and CoEvoKG. Co-first author on both.
Jul 01, 2026	Cognitive Scaffold is published at ACL 2026 (Main).
Jun 11, 2026	WM-DETR is online in Expert Systems with Applications.
May 14, 2026	HiPA-DETR is published in The Visual Computer. Co-first author.
Mar 14, 2026	CACMI appears at AAAI 2026.
Mar 09, 2026	HMR-1 preprint is on arXiv.
Oct 15, 2025	DMTrack preprint is on arXiv, with code.
Aug 28, 2025	Our robot-vision survey is online in Information Fusion.

selected publications

arXiv

CoEvoKG: Co-Evolving Knowledge Graphs with Self-Evolving Search Agents

Zhaoyang Li^∗, Zenghuang Fu^∗, Qiuyuan Ai^∗, and 6 more authors

arXiv preprint arXiv:2608.01904, 2026

arXiv Bib Code

@article{li2026coevokg,
  title = {CoEvoKG: Co-Evolving Knowledge Graphs with Self-Evolving Search Agents},
  author = {Li∗, Zhaoyang and Fu∗, Zenghuang and Ai∗, Qiuyuan and Jiang, Ping and Wu, Haoyu and Wu, Minghui and Zhao, Chenxu and Song, Jie and He, Guannan},
  journal = {arXiv preprint arXiv:2608.01904},
  year = {2026},
}

arXiv

Self-Play Meets Skill Evolution: Self-Evolving Search Agents that Pose, Solve, and Remember

Zenghuang Fu^∗, Zhaoyang Li^∗, Qiuyuan Ai^∗, and 6 more authors

arXiv preprint arXiv:2607.29468, 2026

arXiv Bib Code

@article{fu2026sesa,
  title = {Self-Play Meets Skill Evolution: Self-Evolving Search Agents that Pose, Solve, and Remember},
  author = {Fu∗, Zenghuang and Li∗, Zhaoyang and Ai∗, Qiuyuan and Wu∗, Haoyu and Wu, Minghui and Zhao, Chenxu and Wang, Ante and He, Guannan and Wang, Changwei},
  journal = {arXiv preprint arXiv:2607.29468},
  year = {2026},
}

TVC

Robust detection in complex construction sites: HiPA-DETR with weather-aware and cross-domain generalization

Zenghuang Fu^∗, Jinming Yang^∗, Xiaofeng Han, and 6 more authors

The Visual Computer, 2026

DOI Bib Code

@article{fu2026robust,
  title = {Robust detection in complex construction sites: HiPA-DETR with weather-aware and cross-domain generalization},
  author = {Fu∗, Zenghuang and Yang∗, Jinming and Han, Xiaofeng and Feng, Zhe and Zhang, Muyang and Wang, Changwei and Meng, Weiliang and Zhang, Jiguang and Zhang, Xiaopeng},
  journal = {The Visual Computer},
  volume = {42},
  number = {7},
  pages = {295},
  year = {2026},
  publisher = {Springer Berlin Heidelberg},
  doi = {10.1007/s00371-026-04486-y},
}

ACL

Cognitive Scaffold: From Fluid Context to Crystallized Memory for Long-Horizon DeepResearch Agents

Qiuyuan Ai^∗, Zenghuang Fu^∗, Zhaoyang Li^∗, and 4 more authors

In Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL), 2026

DOI Bib

@inproceedings{ai2026cognitive,
  title = {Cognitive Scaffold: From Fluid Context to Crystallized Memory for Long-Horizon DeepResearch Agents},
  author = {Ai∗, Qiuyuan and Fu∗, Zenghuang and Li∗, Zhaoyang and Jiang, Ping and Wu, Haoyu and Song, Jie and He, Guannan},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (ACL)},
  pages = {25526--25542},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1170}
}

ESWA

WM-DETR: Dual-Branch Wavelet-Mamba and Sparse Attention for Robust Underwater Object Detection

Xiaofeng Han^∗, Shunpeng Chen^∗, Zenghuang Fu, and 7 more authors

Expert Systems with Applications, 2026

DOI Bib Code

@article{han2026wmdetr,
  title = {WM-DETR: Dual-Branch Wavelet-Mamba and Sparse Attention for Robust Underwater Object Detection},
  author = {Han∗, Xiaofeng and Chen∗, Shunpeng and Fu, Zenghuang and Huang, Longzhao and Xu, Shengpeng and Yang, Jinming and Kong, Yixian and Wang, Changwei and Meng, Weiliang and Zhang, Xiaopeng},
  journal = {Expert Systems with Applications},
  pages = {133148},
  year = {2026},
  publisher = {Pergamon},
  doi = {10.1016/j.eswa.2026.133148},
}

AAAI

Explicit temporal-semantic modeling for dense video captioning via context-aware cross-modal interaction

Mingda Jia, Weiliang Meng, Zenghuang Fu, and 7 more authors

In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), 2026

DOI arXiv Bib

@inproceedings{jia2026explicit,
  title = {Explicit temporal-semantic modeling for dense video captioning via context-aware cross-modal interaction},
  author = {Jia, Mingda and Meng, Weiliang and Fu, Zenghuang and Li, Yiheng and Zeng, Qi and Zhang, Yifan and Xin, Ju and Xu, Rongtao and Zhang, Jiguang and Zhang, Xiaopeng},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)},
  volume = {40},
  number = {7},
  pages = {5341--5349},
  year = {2026},
  doi = {10.1609/aaai.v40i7.37450},
}

arXiv

HMR-1: Hierarchical Massage Robot with Vision-Language-Model for Embodied Healthcare

Rongtao Xu^∗, Mingming Yu^∗, Xiaofeng Han^∗, and 7 more authors

arXiv preprint arXiv:2603.08817, 2026

arXiv Bib

@article{xu2026hmr,
  title = {HMR-1: Hierarchical Massage Robot with Vision-Language-Model for Embodied Healthcare},
  author = {Xu∗, Rongtao and Yu∗, Mingming and Han∗, Xiaofeng and Zhang, Yu and Hu, Kaiyi and Feng, Zhe and Fu, Zenghuang and Wang, Changwei and Meng, Weiliang and Zhang, Xiaopeng},
  journal = {arXiv preprint arXiv:2603.08817},
  year = {2026},
}

arXiv

DMTrack: Deformable State-Space Modeling for UAV Multi-Object Tracking with Kalman Fusion and Uncertainty-Aware Association

Zenghuang Fu^∗, Xiaofeng Han^∗, Mingda Jia, and 5 more authors

arXiv preprint arXiv:2510.17860, 2025

arXiv Bib Code

@article{fu2025dmtrack,
  title = {DMTrack: Deformable State-Space Modeling for UAV Multi-Object Tracking with Kalman Fusion and Uncertainty-Aware Association},
  author = {Fu∗, Zenghuang and Han∗, Xiaofeng and Jia, Mingda and Zeng, Qi and Zhang, Muyang and Wang, Changwei and Meng, Weiliang and Zhang, Xiaopeng},
  journal = {arXiv preprint arXiv:2510.17860},
  year = {2025},
}

InfFus

Multimodal fusion and vision-language models: A survey for robot vision

Xiaofeng Han^∗, Shunpeng Chen^∗, Zenghuang Fu, and 9 more authors

Information Fusion, 2025

DOI arXiv Bib Code

@article{han2025multimodal,
  title = {Multimodal fusion and vision-language models: A survey for robot vision},
  author = {Han∗, Xiaofeng and Chen∗, Shunpeng and Fu, Zenghuang and Feng, Zhe and Fan, Lue and An, Dong and Wang, Changwei and Guo, Li and Meng, Weiliang and Zhang, Xiaopeng and Xu, Rongtao and Xu, Shibiao},
  journal = {Information Fusion},
  pages = {103652},
  year = {2025},
  publisher = {Elsevier},
  doi = {10.1016/j.inffus.2025.103652},
}