EUVIMEDEUVIMED entdecken
Crossref

Counterfactual Planning for Generalizable Agents’ Actions

Jiarun Fu, Lizhong Ding, Qiuning Wei, Yuhan Guo, Yurong Cheng, Junyu Zhang

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (35) · 29432-29440

DOI: 10.1609/aaai.v40i35.40184 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Jiarun Fu und 5 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Large language models have revolutionized agent planning by serving as the engine of heuristic guidance. However, LLM-based agents often struggle to generalize across complex environments and to adapt to stochastic feedback arising from environment–action interactions. We propose Counterfactual Planning—a method designed to improve the generalizability and adaptability of agents' actions by inferring causal representations of environmental confounders and performing counterfactual reasoning over planned actions. We formalize the agent planning process as a structural causal model, providing a mathematical formulation for causal analysis of how environmental states influence action generation and how actions affect future state transitions. To support generalizable action planning, we introduce the State Causality Evaluator (SCE), which dynamically infers task-conditioned causal representations from complex environment states; and to enhance adaptability under stochastic feedback, we propose the What-If-Not (WIN) reward, which performs counterfactual interventions to refine actions through causal evaluation. We validate our framework in an open-world environment, where experiments demonstrate improvements in both action generalization and planning adaptability.

Zitieren

APA

Fu, J., Ding, L., Wei, Q., Guo, Y., Cheng, Y., & Zhang, J. (2026). Counterfactual Planning for Generalizable Agents’ Actions. Proceedings of the AAAI Conference on Artificial Intelligence, 40(35), 29432-29440. https://doi.org/10.1609/aaai.v40i35.40184

Vancouver

Fu J, Ding L, Wei Q, Guo Y, Cheng Y, Zhang J. Counterfactual Planning for Generalizable Agents’ Actions. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(35):29432-29440. doi:10.1609/aaai.v40i35.40184

BibTeX
@article{Fu2026,
  author = {Fu, Jiarun and Ding, Lizhong and Wei, Qiuning and Guo, Yuhan and Cheng, Yurong and Zhang, Junyu},
  title = {Counterfactual Planning for Generalizable Agents’ Actions},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {35},
  pages = {29432--29440},
  doi = {10.1609/aaai.v40i35.40184}
}
Kostenlos starten →