# Marathon Environments A set of high-dimensional continuous control environments for use with Unity ML-Agents Toolkit. ![MarathonEnvs](https://github.com/Unity-Technologies/marathon-envs/raw/master/images/MarathonEnvsBanner.gif) ## Preview MarathonEnvs using the **[Web Demo](http://marathonenvs.joebooth.com)** **MarathonEnvs** is a set of high-dimensional continuous control benchmarks using Unity’s native physics simulator, PhysX. MarathonEnvs can be trained using Unity ML-Agents or any OpenAI Gym compatible algorithm. MarathonEnvs may be useful for: * Video Game researchers interested in apply bleeding-edge robotics research into the domain of locomotion and AI for video games. * Academic researchers looking to leverage the strengths of Unity and ML-Agents along with the body of existing research and benchmarks provided by projects such as the [DeepMind Control Suite](https://github.com/deepmind/dm_control), or [OpenAI Mujoco](http://gym.openai.com/envs/#mujoco) environments. *Note: This project is the result of contributions from members of the Unity community (see below) who actively maintain the repository. As such, the contents of this repository are not officially supported by Unity Technologies.* * [Joe Booth (@Sohojoe)](https://github.com/Sohojoe), Twitter - [@iAmVidyaGamer](https://twitter.com/iAmVidyaGamer) * [Vladimir Ivanov (@vivanov879)](https://github.com/vivanov879) **Need Help?** * Open an [issue](https://github.com/Unity-Technologies/marathon-envs/issues) * Join our [Discord server](https://discord.gg/MPEbHPP) * Say hi on Twitter - [@iAmVidyaGamer](https://twitter.com/iAmVidyaGamer) --- ## Environments | **Controller (DReCon)** - Preview | | |:----------|:----------| | ![Controller](https://github.com/Unity-Technologies/marathon-envs/raw/master/images/Controller.gif) |

A controller based agent, inspired by the DReCon paper (link below). The agent learns to follow a simple traditional controller agent and exhibits emergent behavior. **In Preview**

ControllerMarathonMan-v0

Learning from motion capture examples, inspired by the DeepMimic paper (link below). The agent learns the motion capture sequence using a phase value.

MarathonManWalking-v0
MarathonManRunning-v0
arathonManJazzDancing-v0
MarathonManMMAKick-v0
MarathonManPunchingBag-v0
MarathonManBackflip-v0

| ![StyleTransfer](https://github.com/Unity-Technologies/marathon-envs/raw/master/images/StyleTransfer.gif) | | **Procedural Environments** | | | ![Terrain](https://github.com/Unity-Technologies/marathon-envs/raw/master/images/Terrain.gif) |

Procedurally-generated terrains aimed at addressing overfitting in Reinforcement Learning and generalizable skills.

TerrainHopper-v0
TerrainWalker2d-v0
TerrainAnt-v0
TerrainMarathonMan-v0

Classical implementations of Ant, Hopper, Walker-2d, Humanoid

Hopper-v0
Walker2d-v0
Ant-v0
MarathonMan-v0

Sparse reward version of a humanoid learning to walk. The agent recives a single reward at the end of the episode.

MarathonManSparse-v0