A Framework for Scalable Heterogeneous Multi-Agent Adversarial Reinforcement Learning in IsaacLab
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911188557758464 |
|---|---|
| author | Peterson, Isaac Allred, Christopher Morrey, Jacob Harper, Mario |
| author_facet | Peterson, Isaac Allred, Christopher Morrey, Jacob Harper, Mario |
| contents | Multi-Agent Reinforcement Learning (MARL) is central to robotic systems cooperating in dynamic environments. While prior work has focused on these collaborative settings, adversarial interactions are equally critical for real-world applications such as pursuit-evasion, security, and competitive manipulation. In this work, we extend the IsaacLab framework to support scalable training of adversarial policies in high-fidelity physics simulations. We introduce a suite of adversarial MARL environments featuring heterogeneous agents with asymmetric goals and capabilities. Our platform integrates a competitive variant of Heterogeneous Agent Reinforcement Learning with Proximal Policy Optimization (HAPPO), enabling efficient training and evaluation under adversarial dynamics. Experiments across several benchmark scenarios demonstrate the framework's ability to model and train robust policies for morphologically diverse multi-agent competition while maintaining high throughput and simulation realism. Code and benchmarks are available at: https://github.com/DIRECTLab/IsaacLab-HARL . |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_01264 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A Framework for Scalable Heterogeneous Multi-Agent Adversarial Reinforcement Learning in IsaacLab Peterson, Isaac Allred, Christopher Morrey, Jacob Harper, Mario Machine Learning Robotics 68T42 (Primary) 68T40, 68T05 (Secondary) Multi-Agent Reinforcement Learning (MARL) is central to robotic systems cooperating in dynamic environments. While prior work has focused on these collaborative settings, adversarial interactions are equally critical for real-world applications such as pursuit-evasion, security, and competitive manipulation. In this work, we extend the IsaacLab framework to support scalable training of adversarial policies in high-fidelity physics simulations. We introduce a suite of adversarial MARL environments featuring heterogeneous agents with asymmetric goals and capabilities. Our platform integrates a competitive variant of Heterogeneous Agent Reinforcement Learning with Proximal Policy Optimization (HAPPO), enabling efficient training and evaluation under adversarial dynamics. Experiments across several benchmark scenarios demonstrate the framework's ability to model and train robust policies for morphologically diverse multi-agent competition while maintaining high throughput and simulation realism. Code and benchmarks are available at: https://github.com/DIRECTLab/IsaacLab-HARL . |
| title | A Framework for Scalable Heterogeneous Multi-Agent Adversarial Reinforcement Learning in IsaacLab |
| topic | Machine Learning Robotics 68T42 (Primary) 68T40, 68T05 (Secondary) |
| url | https://arxiv.org/abs/2510.01264 |