gym-video-streamer 1.0

Wrapper to live stream OpenAI's gym agents training process from Google Colab to Twitch/YouTube server.

Gym Video Streamer

A Video Streaming Wrapper for OpenAI's Gym Environments

Now you can Live Stream the Agent's learning behavior to Twitch/YouTube from Google Colab while training these Agents.

What’s In This Document

• Installation
• Imports and Usage
• Setup for different type of Gym Envs
• License

Installation

!pip install gym-video-streamer

And if you already have gym-video-streamer then upgrade it by this command.

!pip install --upgrade gym-video-streamer

Imports and Usage

import gym
from gym_video_streamer import SetupVirtualDisplay
from gym_video_streamer import VideoStreamer  # Streaming Wrapper

• Now Setup the Virtual Display (only required for Google Colab):

  SetupVirtualDisplay()
  

• Define your live stream information:

  # stream_info dictionary should be in this format only
  stream_info = {
      "URL": "rtmp://live.twitch.tv/app/", # example of Twitch URL
      "secret": "--- secret here ---"
  }
  

• Initialize the gym env and pass it to the custom wrapper:

  # ---- {For Classic-control gym envs} ----
  env = VideoStreamer(gym.make("CartPole-v1"), stream_info)
  

  *If you don't pass stream_info then it will simply store the video locally in the videos directory.

• Test the setup (running 100 episodes for testing):

  try:
      observation = env.reset()
      i = 0
      while True:
          if i == 100:
              break
  
          env.render()  # important to call render method on env
          action = env.action_space.sample()
          observation, reward, done, info = env.step(action)
  
          if done:
              env.reset()
              i += 1
  except Exception as e:
      print(e)
  finally:
      env.close()
  

Setup for different type of Gym Envs

• Classic control Gym Envs: As we have seen in the above example this type of works without any extra installation/setup.

• Box2D Gym Envs: For this type of envs you need to install the following packages:

  !pip install box2d box2d-py
  

• Atari Gym Envs: Using this type of envs on Google Colab you, need some extra setup to make them working. When I tried, it gave me the following error Exception: ROM is missing for breakout, see https://github.com/openai/atari-py#roms for instructions. So if you know how to setup this env on Colab then do let me know ✌🏻.

• MuJoCo & Robotics Gym Envs: Now for this type of envs, you need to setup the MuJoCo on Colab. And again I haven't done this but I found a resource that will help you Setup Mujoco-py on Linux.

License

Licensed under the MIT License.