MarineGym Documentation¶

MarineGym is an large-scale parallel framework designed for reinforcement learning research on unmanned underwater vehicles (UUVs). It is build upon OmniDrones and Isaac Sim, offering the following features:

Efficiency: Achieve a simulation speed of up to 10^7 steps per second.
Fidelity: Accurately replicate the physical environment, encompassing physical laws, kinematics, and dynamics.
Flexibility: Ensure compatibility with existing RL frameworks and offering user-friendly APIs to facilitate seamless integration and usage.
Evaluation: Assess and contrast various RL strategies through multiple tasks.

If you build on this work, please cite:

@online{chu_2025_MarineGymHighPerformanceReinforcement,
        title = {MarineGym: A High-Performance Reinforcement Learning Platform for Underwater Robotics},
        shorttitle = {MarineGym},
        author = {Chu, Shuguang and Huang, Zebin and Li, Yutong and Lin, Mingwei and Carlucho, Ignacio and Petillot, Yvan R. and Yang, Canjun},
        date = {2025-03-12},
        eprint = {2503.09203},
        eprinttype = {arXiv},
        eprintclass = {cs},
        doi = {10.48550/arXiv.2503.09203},
        pubstate = {prepublished}
      }

Getting Started

Demos

API Reference

API Reference