Elamur Research⌕
☰
home / papers / 2026 / RSMeM: Knowledge-Enhanced Memory Evolution for Remote Sensing Agents with Systematic Evaluation

RSMeM: Knowledge-Enhanced Memory Evolution for Remote Sensing Agents with Systematic Evaluation

Paper · 2026 · Bingxian Wu, Yu Zhang, Zonghao Guo et al.

Summary

Remote sensing (RS) agents built by attaching tools to general-purpose LLMs often fail on geoscience workflows. They mix up domain concepts, call wrong or non-existent tools, and do not learn from these errors on later tasks. Template-based alternatives avoid some errors but require expensive expert curation. RSMeM addresses this with a memory that starts from curated domain knowledge and grows from the agent's own mistakes.

Offline, the authors organise geoscience knowledge into a taxonomy of task cards, refined with expert feedback. Online, each query draws on two sources. Hierarchical Knowledge Grounding retrieves task-level guidance using keywords, task definitions and embeddings. Failure-Aware Experience Refinement critiques flawed tool-use traces and condenses them into reusable constraints for later attempts.

Over repeated rounds on Earth-Bench, this loop improves both tool use and final answers for every tested backbone. The gains cost almost no extra tokens. The approach therefore offers a cheap, plug-and-play way to make Earth-observation agents more reliable.

Key points

  • The knowledge base is organised into 3 domains of application, which divide into 12 subdomains and cover 64 distinct task formulations. Each task card holds a title, a domain definition and execution hints.
  • Retrieval works in two streams: Hierarchical Knowledge Grounding (HKG) supplies task-level guidance, and a Failure-Aware Memory (FAM) supplies instance-level reflections. Both feed a ReAct-style agent equipped with external RS tools.
  • Evaluation uses Earth-Bench with four backbones: DeepSeek-V3.2, Kimi-K2, Qwen3-32B and Qwen3-8B. Stronger models gain more because they diagnose their own errors better.
  • On DeepSeek-V3.2, accuracy rises by 6.07 points (absolute) while adding under 1% extra experience tokens.
  • In a Landsat thermal case study, the agent first treated BT10 brightness temperature as LST. The distilled constraint led it to compute emissivity-corrected LST in the next round, which produced the correct answer of 5 days.
  • The authors note two limitations: early performance depends on the coverage of the knowledge corpus, and only failed traces are currently distilled.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Tsinghua UniversityBeijing · China415 works · 94,028 citations🎓Shanghai Jiao Tong UniversityShanghai · China212 works · 19,544 citations🏛️Chinese Academy of SciencesBeijing · China246 works · 24,952 citations🎓University of Chinese Academy of SciencesBeijing · China188 works · 8,744 citations🎓China University of Geosciences (Beijing)Beijing · China3 works · 33 citations🎓Aerospace Information Research InstituteBeijing · China3 works

Also on the byline: Institute of Geographic Sciences and Natural Resources Research.

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 18,614 worksComputer Vision and Pattern Recognition · 3,486 worksEngineering · 188 worksMedia Technology · 42 worksEarth and Planetary Sciences · 4 worksAtmospheric Science · 2 works

Cite

Cite as Bingxian Wu, Yu Zhang, Zonghao Guo, Tang Liu, Chen Qian, Yuxiang Lu, Xingbo Du, Yanghao Li, Yidan Zhang, Chi Chen, Ling Yao, Chenghu Zhou and Maosong Sun. 2026. RSMeM: Knowledge-Enhanced Memory Evolution for Remote Sensing Agents with Systematic Evaluation. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.acl-long.1519

@inproceedings{wu2026rsmem,
  title = {{RSMeM: Knowledge-Enhanced Memory Evolution for Remote Sensing Agents with Systematic Evaluation}},
  author = {Bingxian Wu, Yu Zhang, Zonghao Guo, Tang Liu, Chen Qian, Yuxiang Lu, Xingbo Du, Yanghao Li, Yidan Zhang, Chi Chen, Ling Yao, Chenghu Zhou and Maosong Sun},
  booktitle = {Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)},
  year = {2026},
  doi = {10.18653/v1/2026.acl-long.1519},
  url = {https://doi.org/10.18653/v1/2026.acl-long.1519}
}

References

36 references, 8 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Yuxuan Cai, Yipeng Hao, Jie Zhou, Hang Yan, Zhikai Lei, Rui Zhen, Zhenhua Han, Yutao Yang, Junsong Li, Qianjun Pan, Tianyu Huai, Qin Chen, Xin Li, Kai Chen, Bo Zhang, Xipeng Qiu, and Liang He. 2025. Building self-evolving agents via experience-driven lifelong learning: A framework and benchmark. Pre… — [arXiv:2508.19005](https://arxiv.org/abs/2508.19005) find
  2. Hongruixuan Chen, Chen Wu, Bo Du, Liangpei Zhang, and Le Wang. 2020. Change detection in multisource vhr images via deep siamese convolutional multiple-layers recurrent neural network. IEEE Transactions on Geoscience and Remote Sensing, 58(4):2848– 2864. find
  3. Zhengchao Chen, Haoran Wang, Jing Yao, Pedram Ghamisi, Jun Zhou, Peter M. Atkinson, and Bing Zhang. 2025. Cangling-knowflow: A unified knowledge-and-flow-fused agent for comprehensive remote sensing applications. Preprint, arXiv:2512.15231. — [arXiv:2512.15231](https://arxiv.org/abs/2512.15231) find
  4. Prateek Chhikara, Dev Khant, Saket Aryan, Taranjeet Singh, and Deshraj Yadav. 2025. Mem0: Building production-ready ai agents with scalable long-term memory. Preprint, arXiv:2504.19413.
  5. DeepSeek-AI, Aixin Liu, Bei Feng, Bing Xue, Bingxuan Wang, Bochao Wu, Chengda Lu, Chenggang Zhao, Chengqi Deng, Chenyu Zhang, Chong Ruan, Damai Dai, Daya Guo, Dejian Yang, Deli Chen, Dongjie Ji, Erhang Li, Fangyun Lin, Fucong Dai, and 181 others. 2025. Deepseek-v3 technical report. Preprint, arXiv:2… find
  6. Yue Fan, Xiaojian Ma, Rujie Wu, Yuntao Du, Jiaqi Li, Zhi Gao, and Qing Li. 2024. Videoagent: A memory-augmented multimodal agent for video understanding. Preprint, arXiv:2403.11481. — [arXiv:2403.11481](https://arxiv.org/abs/2403.11481) find
  7. Runnan Fang, Yuan Liang, Xiaobin Wang, Jialong Wu, Shuofei Qiao, Pengjun Xie, Fei Huang, Huajun Chen, and Ningyu Zhang. 2025. Memp: Exploring agent procedural memory. Preprint, arXiv:2508.06433. find
  8. Peilin Feng, Zhutao Lv, Junyan Ye, Xiaolei Wang, Xinjie Huo, Jinhua Yu, Wanghan Xu, Wenlong Zhang, Lei Bai, Conghui He, and Weijia Li. 2025. Earth-agent: Unlocking the full landscape of earth observation with agents. Preprint, arXiv:2509.23141. — [arXiv:2509.23141](https://arxiv.org/abs/2509.23141) find
  9. Haonan Guo, Xin Su, Chen Wu, Bo Du, Liangpei Zhang, and Deren Li. 2024. Remote sensing chatgpt: Solving remote sensing tasks with chatgpt and visual models. Preprint, arXiv:2401.09083. — [arXiv:2401.09083](https://arxiv.org/abs/2401.09083) find
  10. Peng Luo, Xiayin Lou, Yu Zheng, Zhuo Zheng, and Stefano Ermon. 2025. Geoevolve: Automating geospatial model discovery via multi-agent large language models. Preprint, arXiv:2509.21593. find
  11. Jingjing Ma, Wei Jiang, Xu Tang, Xiangrong Zhang, Fang Liu, and Licheng Jiao. 2025. Multiscale sparse cross-attention network for remote sensing scene classification. IEEE Transactions on Geoscience and Remote Sensing, 63:1–16. find
  12. Mengkang Hu, Tianxing Chen, Qiguang Chen, Yao Mu, Wenqi Shao, and Ping Luo. 2025. HiAgent: Hierarchical working memory management for solving long-horizon agent tasks with large language model. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long…
  13. Yuchi Ma, Shuo Chen, Stefano Ermon, and David B. Lobell. 2024. Transfer learning in environmental remote sensing. Remote Sensing of Environment, 301:113924. find
  14. Dong Huang, Jie M. Zhang, Michael Luck, Qingwen Bu, Yuhao Qing, and Heming Cui. 2024. Agentcoder: Multi-agent-based code generation with iterative testing and optimisation. Preprint, arXiv:2312.13010.
  15. Akashah Shabbir, Muhammad Akhtar Munir, Akshay Dudhane, Muhammad Umer Sheikh, Muhammad Haris Khan, Paolo Fraccaro, Juan Bernabe Moreno, Fahad Shahbaz Khan, and Salman Khan. 2025. Thinkgeo: Evaluating tool-augmented agents for remote sensing tasks. Preprint, arXiv:2505.23752. — [arXiv:2505.23752](https://arxiv.org/abs/2505.23752) find
  16. Chia Hsiang Kao, Wenting Zhao, Shreelekha Re-vankar, Samuel Speas, Snehal Bhagat, Rajeev Datta, Cheng Perng Phoo, Utkarsh Mall, Carl Vondrick, Kavita Bala, and Bharath Hariharan. 2025a. Towards llm agents for earth observation. Preprint, arXiv:2504.12110. find
  17. Chia Hsiang Kao, Wenting Zhao, Shreelekha Re-vankar, Samuel Speas, Snehal Bhagat, Rajeev Datta, Cheng Perng Phoo, Utkarsh Mall, Carl Vondrick, Kavita Bala, and Bharath Hariharan. 2025b. Towards llm agents for earth observation: The univearth dataset. arXiv preprint. find
  18. Zhengjing Ma and Gang Mei. 2021. Deep learning for geological hazards analysis: Data, models, applications, and opportunities. Earth-Science Reviews, 223:103858. find
  19. Noah Shinn, Federico Cassano, Edward Berman, Ashwin Gopinath, Karthik Narasimhan, and Shunyu Yao. 2023. Reflexion: Language agents with verbal reinforcement learning. Preprint, arXiv:2303.11366.
  20. Yu Su, Diyi Yang, Shunyu Yao, and Tao Yu. 2024. Language agents: Foundations, prospects, and risks. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Tutorial Abstracts, pages 17–24, Miami, Florida, USA. Association for Computational Linguistics. find
  21. Yubin Kim, Ken Gu, Chanwoo Park, Chunjong Park, Samuel Schmidgall, A. Ali Heydari, Yao Yan, Zhihan Zhang, Yuchen Zhuang, Mark Malhotra, Paul Pu Liang, Hae Won Park, Yuzhe Yang, Xuhai Xu, Yilun Du, Shwetak Patel, Tim Althoff, Daniel McDuff, and Xin Liu. 2025. Towards a science of scaling agent system… — [arXiv:2512.08296](https://arxiv.org/abs/2512.08296) find
  22. Jingkai Sun, Qiang Zhang, Yiqun Duan, Xiaoyang Jiang, Chong Cheng, and Renjing Xu. 2024. Prompt, plan, perform: Llm-based humanoid control via quantized imitation learning. In 2024 IEEE International Conference on Robotics and Automation (ICRA), pages 16236–16242. find
  23. Maja Kucharczyk and Chris H. Hugenholtz. 2021. Remote sensing of natural hazard-related disasters with small drones: Global trends, biases, and research opportunities. Remote Sensing of Environment, 264:112577. find
  24. Xiaoyu Tan, Bin Li, Xihe Qiu, Chao Qu, Wei Chu, Yinghui Xu, and Yuan Qi. 2025. Meta-agent-workflow: Streamlining tool usage in llms through workflow construction, retrieval, and refinement. New York, NY, USA. Association for Computing Machinery. find
  25. Kaiyu Li, Jiayu Wang, Zhi Wang, Hui Qiao, Weizhan Zhang, Deyu Meng, and Xiangyong Cao. 2025. Designing domain-specific agents via hierarchical task abstraction mechanism. Preprint, arXiv:2511.17198. — [arXiv:2511.17198](https://arxiv.org/abs/2511.17198) find
  26. Chenyang Liu, Keyan Chen, Haotian Zhang, Zipeng Qi, Zhengxia Zou, and Zhenwei Shi. 2024. Change-agent: Toward interactive comprehensive remote sensing change interpretation and analysis. IEEE Transactions on Geoscience and Remote Sensing, 62:1–16. find
  27. Xiangru Tang, Tianrui Qin, Tianhao Peng, Ziyang Zhou, Daniel Shao, Tingting Du, Xinming Wei, Peng Xia, Fang Wu, He Zhu, Ge Zhang, Jiaheng Liu, Xingyao Wang, Sirui Hong, Chenglin Wu, Hao Cheng, Chi Wang, and Wangchunshu Zhou. 2025. Agent kb: Leveraging cross-domain experience for agentic problem solv… — [arXiv:2507.06229](https://arxiv.org/abs/2507.06229) find
  28. Kimi Team, Yifan Bai, Yiping Bao, Guanduo Chen, Jiahao Chen, Ningxin Chen, Ruijue Chen, Yanru Chen, Yuankun Chen, Yutian Chen, Zhuofu Chen, Jialei Cui, Hao Ding, Mengnan Dong, Angang Du, Chenzhuang Du, Dikang Du, Yulun Du, Yu Fan, and 150 others. 2025. Kimi k2: Open agentic intelligence. Preprint, a…
  29. Junyang Wang, Haiyang Xu, Haitao Jia, Xi Zhang, Ming Yan, Weizhou Shen, Ji Zhang, Fei Huang, and Jitao Sang. 2024. Mobile-agent-v2: Mobile device operation assistant with effective navigation via multi-agent collaboration. In Advances in Neural Information Processing Systems, volume 37, pages 2686–… find
  30. Wenjia Xu, Zijian Yu, Boyang Mu, Zhiwei Wei, Yuan-ben Zhang, Guangzuo Li, and Mugen Peng. 2025a. Rs-agent: Automating remote sensing tasks through intelligent agent. Preprint, arXiv:2406.07089. — [arXiv:2406.07089](https://arxiv.org/abs/2406.07089) find
  31. Wujiang Xu, Zujie Liang, Kai Mei, Hang Gao, Juntao Tan, and Yongfeng Zhang. 2025b. A-mem: Agentic memory for llm agents. Preprint, arXiv:2502.12110.
  32. An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, and 41 others. 2025. Qwen3 technical report. Preprint, arXiv:2505.09388.
  33. Qiangqiang Yuan, Huanfeng Shen, Tongwen Li, Zhiwei Li, Shuwen Li, Yun Jiang, Hongzhang Xu, Weiwei Tan, Qianqian Yang, Jiwen Wang, Jianhao Gao, and Liangpei Zhang. 2020. Deep learning in environmental remote sensing: Achievements and challenges. Remote Sensing of Environment, 241:111716. find
  34. Guibin Zhang, Haotian Ren, Chong Zhan, Zhenhong Zhou, Junhao Wang, He Zhu, Wangchunshu Zhou, and Shuicheng Yan. 2025a. Memevolve: Meta-evolution of agent memory systems. Preprint, arXiv:2512.18746. — [arXiv:2512.18746](https://arxiv.org/abs/2512.18746) find
  35. Kai Zhang, Xiangchao Chen, Bo Liu, Tianci Xue, Zeyi Liao, Zhihan Liu, Xiyao Wang, Yuting Ning, Zhaorun Chen, Xiaohan Fu, Jian Xie, Yuxuan Sun, Boyu Gou, Qi Qi, Zihang Meng, Jianwei Yang, Ning Zhang, Xian Li, Ashish Shah, and 11 others. 2025b. Agent learning via early experience. Preprint, arXiv:2510… find
  36. Wenyuan Zhang, Xinghua Zhang, Haiyang Yu, Shuaiyi Nie, Bingli Wu, Juwei Yue, Tingwen Liu, and Yongbin Li. 2026. Expseek: Self-triggered experience seeking for web agents. Preprint, arXiv:2601.08605.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wu-2026-rsmem-knowledge-enhanced-memory-evolution/