Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imagination

Duy Phuong Nguyen, Kai-Chieh Hsu, Wenhao Yu, Jie Tan, Jaime Fernández Fisac
Proceedings of The 8th Conference on Robot Learning, PMLR 270:387-407, 2025.

Abstract

Despite the impressive recent advances in learning-based robot control, ensuring robustness to out-of-distribution conditions remains an open challenge. Safety filters can, in principle, keep arbitrary control policies from incurring catastrophic failures by overriding unsafe actions, but existing solutions for complex (e.g., legged) robot dynamics do not span the full motion envelope and instead rely on local, reduced-order models. These filters tend to overly restrict agility and can still fail when perturbed away from nominal conditions. This paper presents the gameplay filter, a new class of predictive safety filter that continually plays out hypothetical matches between its simulation-trained safety strategy and a virtual adversary co-trained to invoke worst-case events and sim-to-real error, and precludes actions that would cause failures down the line. We demonstrate the scalability and robustness of the approach with a first-of-its-kind full-order safety filter for (36-D) quadrupedal dynamics. Physical experiments on two different quadruped platforms demonstrate the superior zero-shot effectiveness of the gameplay filter under large perturbations such as tugging and unmodeled terrain. Experiment videos and open-source software are available online: https://saferobotics.org/research/gameplay-filter

Cite this Paper

BibTeX
@InProceedings{pmlr-v270-nguyen25a, title = {Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imagination}, author = {Nguyen, Duy Phuong and Hsu, Kai-Chieh and Yu, Wenhao and Tan, Jie and Fisac, Jaime Fern\'andez}, booktitle = {Proceedings of The 8th Conference on Robot Learning}, pages = {387--407}, year = {2025}, editor = {Agrawal, Pulkit and Kroemer, Oliver and Burgard, Wolfram}, volume = {270}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v270/main/assets/nguyen25a/nguyen25a.pdf}, url = {https://proceedings.mlr.press/v270/nguyen25a.html}, abstract = {Despite the impressive recent advances in learning-based robot control, ensuring robustness to out-of-distribution conditions remains an open challenge. Safety filters can, in principle, keep arbitrary control policies from incurring catastrophic failures by overriding unsafe actions, but existing solutions for complex (e.g., legged) robot dynamics do not span the full motion envelope and instead rely on local, reduced-order models. These filters tend to overly restrict agility and can still fail when perturbed away from nominal conditions. This paper presents the gameplay filter, a new class of predictive safety filter that continually plays out hypothetical matches between its simulation-trained safety strategy and a virtual adversary co-trained to invoke worst-case events and sim-to-real error, and precludes actions that would cause failures down the line. We demonstrate the scalability and robustness of the approach with a first-of-its-kind full-order safety filter for (36-D) quadrupedal dynamics. Physical experiments on two different quadruped platforms demonstrate the superior zero-shot effectiveness of the gameplay filter under large perturbations such as tugging and unmodeled terrain. Experiment videos and open-source software are available online: https://saferobotics.org/research/gameplay-filter} }
Endnote
%0 Conference Paper %T Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imagination %A Duy Phuong Nguyen %A Kai-Chieh Hsu %A Wenhao Yu %A Jie Tan %A Jaime Fernández Fisac %B Proceedings of The 8th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2025 %E Pulkit Agrawal %E Oliver Kroemer %E Wolfram Burgard %F pmlr-v270-nguyen25a %I PMLR %P 387--407 %U https://proceedings.mlr.press/v270/nguyen25a.html %V 270 %X Despite the impressive recent advances in learning-based robot control, ensuring robustness to out-of-distribution conditions remains an open challenge. Safety filters can, in principle, keep arbitrary control policies from incurring catastrophic failures by overriding unsafe actions, but existing solutions for complex (e.g., legged) robot dynamics do not span the full motion envelope and instead rely on local, reduced-order models. These filters tend to overly restrict agility and can still fail when perturbed away from nominal conditions. This paper presents the gameplay filter, a new class of predictive safety filter that continually plays out hypothetical matches between its simulation-trained safety strategy and a virtual adversary co-trained to invoke worst-case events and sim-to-real error, and precludes actions that would cause failures down the line. We demonstrate the scalability and robustness of the approach with a first-of-its-kind full-order safety filter for (36-D) quadrupedal dynamics. Physical experiments on two different quadruped platforms demonstrate the superior zero-shot effectiveness of the gameplay filter under large perturbations such as tugging and unmodeled terrain. Experiment videos and open-source software are available online: https://saferobotics.org/research/gameplay-filter
APA
Nguyen, D.P., Hsu, K., Yu, W., Tan, J. & Fisac, J.F.. (2025). Gameplay Filters: Robust Zero-Shot Safety through Adversarial Imagination. Proceedings of The 8th Conference on Robot Learning, in Proceedings of Machine Learning Research 270:387-407 Available from https://proceedings.mlr.press/v270/nguyen25a.html.

Related Material