Elamur Research⌕
☰
home / papers / 2026 / Communication-Efficient Desire Alignment for Proactive Embodied Human–Agent Interaction

Communication-Efficient Desire Alignment for Proactive Embodied Human–Agent Interaction

Paper · 2026 · Yuanfei Wang, Xinju Huang, Fangwei Zhong et al.

Summary

Household assistants serve the same people again and again, and users do not want to spell out exact instructions every time. This paper asks how an embodied agent can learn an unfamiliar user's preferences quickly while keeping dialogue to a minimum. The authors build HA-Desire, a 3D home-assistance simulator based on VirtualHome. In it, an LLM-driven proxy user holds hidden value attributes, such as being thirsty or wanting something refreshing. These attributes determine the user's goals, but the user gives only vague, preference-style hints in natural language. Using this testbed, the authors introduce an LLM-based agent called FAMER (Fast Adaptation via MEntal Reasoning), built from three components: inferring what the user desires, filtering out needless questions through reflection, and a goal-relevant memory that persists from one episode to the next. On household tasks such as snack preparation and table setting, FAMER beats baselines including CoELA, ProAgent and MHP in both task success and communication efficiency. Results stay stable across different backbone LLMs, and a study with real participants confirms the advantage. The work points toward proactive assistants that need fewer explicit commands.

Key points

  • HA-Desire environment: six home layouts, roughly 80 objects per room, drawn from more than 110 object classes. The proxy user samples latent values and then picks goals from a potential goal set.
  • FAMER memory: it stores goal-related details, such as object locations, across episodes. This lets it outperform baselines even when the user states the goals outright.
  • Episode length on Snack-M: no method scores anything with step limits under 30. FAMER reaches a perfect score in the first episode once the limit exceeds 90 steps. Trial-and-error baselines (ProAgent, MHP) can get worse as episodes grow longer, because they collect wrong-goal penalties.
  • Human study: 8 participants, 192 interaction episodes, ratings on a 7-point Likert scale. FAMER scored about 5.6 for satisfaction, 5.9 for helpfulness and 5.6 for communication efficiency. CoELA scored about 4.4, 4.7 and 4.2, and the ablated FAMER variants also scored lower.
  • Limitations: evaluation is simulation-only, with no real-robot deployment, and the system may be vulnerable to manipulated interaction signals.

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

🎓Peking UniversityBeijing · China244 works · 39,359 citations

 Fields, topics and concepts

📚 Shelves 8The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksSocial Psychology · 306 worksEngineering · 133 worksControl and Systems Engineering · 45 works

Cite

Cite as Yuanfei Wang, Xinju Huang, Fangwei Zhong, Yaodong Yang, Yizhou Wang, Yuanpei Chen and Hao Dong. 2026. Communication-Efficient Desire Alignment for Proactive Embodied Human–Agent Interaction. 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.641

@inproceedings{wang2026communication,
  title = {{Communication-Efficient Desire Alignment for Proactive Embodied Human–Agent Interaction}},
  author = {Yuanfei Wang, Xinju Huang, Fangwei Zhong, Yaodong Yang, Yizhou Wang, Yuanpei Chen and Hao Dong},
  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.641},
  url = {https://doi.org/10.18653/v1/2026.acl-long.641}
}

References

43 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Stefano V Albrecht and Subramanian Ramamoorthy. 2015. A game-theoretic model and best-response learning method for ad hoc coordination in multiagent systems. arXiv preprint arXiv:1506.01170. — [arXiv:1506.01170](https://arxiv.org/abs/1506.01170) find
  2. Nolan Bard, Jakob N Foerster, Sarath Chandar, Neil Burch, Marc Lanctot, H Francis Song, Emilio Parisotto, Vincent Dumoulin, Subhodeep Moitra, Edward Hughes, and 1 others. 2020. The hanabi challenge: A new frontier for ai research. Artificial Intelligence, 280:103216. find
  3. James R Bettman, Mary Frances Luce, and John W Payne. 1998. Constructive consumer choice processes. Journal of consumer research, 25(3):187– 217. find
  4. Micah Carroll, Rohin Shah, Mark K Ho, Tom Griffiths, Sanjit Seshia, Pieter Abbeel, and Anca Dragan. 2019. On the utility of learning about humans for human-ai coordination. Advances in Neural Information Processing Systems, 32. find
  5. Shuo Chen, Ewa Andrejczuk, Zhiguang Cao, and Jie Zhang. 2020. Aateam: Achieving the ad hoc teamwork by employing the attention mechanism. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 7095–7102. find
  6. Herbert H Clark and Susan E Brennan. 1991. Grounding in communication. find
  7. Brandon Cui, Hengyuan Hu, Luis Pineda, and Jakob Foerster. 2021. K-level reasoning for zero-shot coordination in hanabi. Advances in Neural Information Processing Systems, 34:8215–8228. find
  8. Josef Dai, Xuehai Pan, Ruiyang Sun, Jiaming Ji, Xinbo Xu, Mickel Liu, Yizhou Wang, and Yaodong Yang. 2023. Safe rlhf: Safe reinforcement learning from human feedback. arXiv preprint arXiv:2310.12773.
  9. Weihua Du, Qiushi Lyu, Jiaming Shan, Zhenting Qi, Hongxin Zhang, Sunli Chen, Andi Peng, Tianmin Shu, Kwonjoon Lee, Behzad Dariush, and 1 others. 2024. Constrained human-ai cooperation: An inclusive embodied social intelligence challenge. In Thirty-eighth Conference on Neural Information Processing S… find
  10. Jaime F Fisac, Monica A Gates, Jessica B Hamrick, Chang Liu, Dylan Hadfield-Menell, Malayandi Palaniappan, Dhruv Malik, S Shankar Sastry, Thomas L Griffiths, and Anca D Dragan. 2020. Pragmatic-pedagogic value alignment. In Robotics Research: the 18th International Symposium ISRR, pages 49–57. Spring… find
  11. Hitesh Goel and Hao Zhu. 2025. Lifelong sotopia: Evaluating social intelligence of language agents over lifelong social interactions. arXiv preprint arXiv:2506.12666. — [arXiv:2506.12666](https://arxiv.org/abs/2506.12666) find
  12. John R Hauser and Birger Wernerfelt. 1990. An evaluation cost model of consideration sets. Journal of consumer research, 16(4):393–408. find
  13. Kaiming He, Georgia Gkioxari, Piotr Dollár, and Ross Girshick. 2017. Mask r-cnn. In Proceedings of the IEEE International Conference on Computer Vision, pages 2961–2969. find
  14. Laura M Hiatt, Cody Narber, Esube Bekele, Sangeet S Khemlani, and J Gregory Trafton. 2017. Human modeling for human–robot collaboration. The International Journal of Robotics Research, 36(5-7):580– 596. find
  15. Hengyuan Hu, Adam Lerer, Brandon Cui, Luis Pineda, Noam Brown, and Jakob Foerster. 2021. Off-belief learning. In International Conference on Machine Learning, pages 4369–4379. PMLR. find
  16. Hengyuan Hu, Adam Lerer, Alex Peysakhovich, and Jakob Foerster. 2020. “other-play” for zero-shot coordination. In International Conference on Machine Learning, pages 4399–4410. PMLR. find
  17. Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, and 1 others. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276.
  18. Jiaming Ji, Tianyi Qiu, Boyuan Chen, Borong Zhang, Hantao Lou, Kaile Wang, Yawen Duan, Zhonghao He, Jiayi Zhou, Zhaowei Zhang, and 1 others. 2023. Ai alignment: A comprehensive survey. arXiv preprint arXiv:2310.19852.
  19. Guohao Li, Hasan Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem. 2023. Camel: Communicative agents for "mind" exploration of large language model society. Advances in Neural Information Processing Systems, 36:51991–52008.
  20. Jie Liu, Pan Zhou, Yingjun Du, Ah-Hwee Tan, Cees GM Snoek, Jan-Jakob Sonke, and Efstratios Gavves. 2025. Capo: Cooperative plan optimization for efficient embodied multi-agent cooperation. In International Conference on Learning Representations. find
  21. Andrei Lupu, Brandon Cui, Hengyuan Hu, and Jakob Foerster. 2021. Trajectory diversity for zero-shot coordination. In International Conference on Machine Learning, pages 7204–7213. PMLR. find
  22. Long Ma, Yuanfei Wang, Fangwei Zhong, Song-Chun Zhu, and Yizhou Wang. 2024. Fast peer adaptation with context-aware exploration. In International Conference on Machine Learning. find
  23. Reuth Mirsky, William Macke, Andy Wang, Harel Yedidsion, and Peter Stone. 2020. A penny for your thoughts: The value of communication in ad hoc teamwork. International Joint Conference on Artificial Intelligence. find
  24. Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, and 1 others. 2022. Training language models to follow instructions with human feedback. Advances in Neural Information Processing Systems, 35:27730–27744.
  25. Joon Sung Park, Joseph O’Brien, Carrie Jun Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. 2023. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 36th annual acm symposium on user interface software and technology, pages 1–22. find
  26. Joon Sung Park, Carolyn Q Zou, Aaron Shaw, Benjamin Mako Hill, Carrie Cai, Meredith Ringel Morris, Robb Willer, Percy Liang, and Michael S Bernstein. 2024. Generative agent simulations of 1,000 people. arXiv preprint arXiv:2411.10109. find
  27. Jinghua Piao, Yuwei Yan, Jun Zhang, Nian Li, Junbo Yan, Xiaochong Lan, Zhihong Lu, Zhiheng Zheng, Jing Yi Wang, Di Zhou, and 1 others. 2025. Agentsociety: Large-scale simulation of llm-driven generative agents advances understanding of human behaviors and society. find
  28. Yuanfei Wang, Fangwei Zhong, Jing Xu, and Yizhou Wang. 2022. Tom2c: Target-oriented multi-agent communication and cooperation with theory of mind. In International Conference on Learning Representations. find
  29. Wendy Wood, Asaf Mazar, and David T Neal. 2022. Habits and goals in human behavior: Separate but interacting systems. Perspectives on Psychological Science, 17(2):590–605. find
  30. Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao. 2023. React: Synergizing reasoning and acting in language models. In International Conference on Learning Representations (ICLR).
  31. Xavier Puig, Kevin Ra, Marko Boben, Jiaman Li, Tingwu Wang, Sanja Fidler, and Antonio Torralba. 2018. Virtualhome: Simulating household activities via programs. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 8494–8502. find
  32. Luyao Yuan, Xiaofeng Gao, Zilong Zheng, Mark Edmonds, Ying Nian Wu, Federico Rossano, Hongjing Lu, Yixin Zhu, and Song-Chun Zhu. 2022. In situ bidirectional human-robot value alignment. Science Robotics, 7(68):eabm4183. find
  33. Xavier Puig, Tianmin Shu, Shuang Li, Zilin Wang, Yuan-Hong Liao, Joshua B Tenenbaum, Sanja Fidler, and Antonio Torralba. 2021. Watch-and-help: A challenge for social perception and human-ai collaboration. In International Conference on Learning Representations. find
  34. Ceyao Zhang, Kaijie Yang, Siyi Hu, Zihao Wang, Guanghe Li, Yihang Sun, Cheng Zhang, Zhaowei Zhang, Anji Liu, Song-Chun Zhu, and 1 others. 2024a. Proagent: building proactive cooperative agents with large language models. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pa… find
  35. Neil Rabinowitz, Frank Perbet, Francis Song, Chiyuan Zhang, SM Ali Eslami, and Matthew Botvinick. 2018. Machine theory of mind. In International Conference on Machine Learning, pages 4218–4227. PMLR. find
  36. Muhammad A Rahman, Niklas Hopner, Filippos Christianos, and Stefano V Albrecht. 2021. Towards open ad hoc teamwork using graph-based policy learning. In International Conference on Machine Learning, pages 8776–8786. PMLR. find
  37. Manish Chandra Reddy Ravula. 2019. Ad-hoc teamwork with behavior-switching agents. In International Joint Conferences on Artificial Intelligence. find
  38. Roger C Schank and Robert P Abelson. 1975. Scripts, plans, and knowledge. Yale University New Haven, CT. find
  39. Peter Stone, Gal Kaminka, Sarit Kraus, and Jeffrey Rosenschein. 2010. Ad hoc autonomous agent teams: Collaboration without pre-coordination. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 24, pages 1504–1509. find
  40. DJ Strouse, Kevin McKee, Matt Botvinick, Edward Hughes, and Richard Everett. 2021. Collaborating with humans without human data. Advances in Neural Information Processing Systems, 34:14502– 14515. find
  41. Yiding Wang, Yuxuan Chen, Fangwei Zhong, Long Ma, and Yizhou Wang. 2025. Simulating human-like daily activities with desire-driven autonomy. In International Conference on Learning Representations. find
  42. Hongxin Zhang, Weihua Du, Jiaming Shan, Qinhong Zhou, Yilun Du, Joshua B Tenenbaum, Tianmin Shu, and Chuang Gan. 2024b. Building cooperative embodied agents modularly with large language models. In International Conference on Learning Representations. find
  43. Xuhui Zhou, Hao Zhu, Leena Mathur, Ruohong Zhang, Haofei Yu, Zhengyang Qi, Louis-Philippe Morency, Yonatan Bisk, Daniel Fried, Graham Neubig, and 1 others. 2023. Sotopia: Interactive evaluation for social intelligence in language agents. arXiv preprint arXiv:2310.11667.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/wang-2026-communication-efficient-desire-alignment-proactive/