OpenAI Gym Environment of the Chrome T-Rex Game

A pygame based port of the Chrome T-Rex Game as an OpenAI Gym Environment.

Important info

You can change the FPS of the game by adjusting the env.FPS value. By default, it is at 60.

Action Space = [0, 1, 2]

0 : No action

1 : Duck

2 : Jump

You can install this from PYPI:

pip3 install gym-dino

You can import it as:

import gym_dino

Small backstory

I am currently working on some reinforcement learning research problems and really wanted a simple yet effective environment to train an agent on. I realised that the Chrome game ain't bad at all especially in this scenario.

So have fun, any help to further develop on this is welcome.

Right now, only render, step, reset and close function have been implemented.

Special thanks to Shivam Shekar's implementation of the original game using pygame.