Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel

Carlota Parés Morlans, Michelle Yi, Claire Chen, Sarah A Wu, Rika Antonova, Tobias Gerstenberg, Jeannette Bohg
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:44874-44886, 2025.

Abstract

Tasks that involve complex interactions between objects with unknown dynamics make planning before execution difficult. These tasks require agents to iteratively improve their actions after actively exploring causes and effects in the environment. For these type of tasks, we propose Causal-PIK, a method that leverages Bayesian optimization to reason about causal interactions via a Physics-Informed Kernel to help guide efficient search for the best next action. Experimental results on Virtual Tools and PHYRE physical reasoning benchmarks show that Causal-PIK outperforms state-of-the-art results, requiring fewer actions to reach the goal. We also compare Causal-PIK to human studies, including results from a new user study we conducted on the PHYRE benchmark. We find that Causal-PIK remains competitive on tasks that are very challenging, even for human problem-solvers.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-morlans25a, title = {Causal-{PIK}: Causality-based Physical Reasoning with a Physics-Informed Kernel}, author = {Morlans, Carlota Par\'{e}s and Yi, Michelle and Chen, Claire and Wu, Sarah A and Antonova, Rika and Gerstenberg, Tobias and Bohg, Jeannette}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {44874--44886}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/morlans25a/morlans25a.pdf}, url = {https://proceedings.mlr.press/v267/morlans25a.html}, abstract = {Tasks that involve complex interactions between objects with unknown dynamics make planning before execution difficult. These tasks require agents to iteratively improve their actions after actively exploring causes and effects in the environment. For these type of tasks, we propose Causal-PIK, a method that leverages Bayesian optimization to reason about causal interactions via a Physics-Informed Kernel to help guide efficient search for the best next action. Experimental results on Virtual Tools and PHYRE physical reasoning benchmarks show that Causal-PIK outperforms state-of-the-art results, requiring fewer actions to reach the goal. We also compare Causal-PIK to human studies, including results from a new user study we conducted on the PHYRE benchmark. We find that Causal-PIK remains competitive on tasks that are very challenging, even for human problem-solvers.} }
Endnote
%0 Conference Paper %T Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel %A Carlota Parés Morlans %A Michelle Yi %A Claire Chen %A Sarah A Wu %A Rika Antonova %A Tobias Gerstenberg %A Jeannette Bohg %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-morlans25a %I PMLR %P 44874--44886 %U https://proceedings.mlr.press/v267/morlans25a.html %V 267 %X Tasks that involve complex interactions between objects with unknown dynamics make planning before execution difficult. These tasks require agents to iteratively improve their actions after actively exploring causes and effects in the environment. For these type of tasks, we propose Causal-PIK, a method that leverages Bayesian optimization to reason about causal interactions via a Physics-Informed Kernel to help guide efficient search for the best next action. Experimental results on Virtual Tools and PHYRE physical reasoning benchmarks show that Causal-PIK outperforms state-of-the-art results, requiring fewer actions to reach the goal. We also compare Causal-PIK to human studies, including results from a new user study we conducted on the PHYRE benchmark. We find that Causal-PIK remains competitive on tasks that are very challenging, even for human problem-solvers.
APA
Morlans, C.P., Yi, M., Chen, C., Wu, S.A., Antonova, R., Gerstenberg, T. & Bohg, J.. (2025). Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:44874-44886 Available from https://proceedings.mlr.press/v267/morlans25a.html.

Related Material