I want to play with this but rather not spend the time to setup a vm and go through the whole pass-through dance or setup a dev env on windows.
I want to play with this but rather not spend the time to setup a vm and go through the whole pass-through dance or setup a dev env on windows.
For the Atari/car racing games you get direct pixel output of the game so using a neural network for image processing is required. But other games have a much simpler environment where you can focus on reinforcement learning.
There's even classic doom. https://gym.openai.com/envs#doom
Trigger Rally: This is available as a nodejs/javascript version (arguably the most advanced):
https://github.com/CodeArtemis/TriggerRally
There is also a native Linux version (check your repo - also here: http://trigger-rally.sourceforge.net/).
There are other racing game simulators available for Linux - TORCS (The Open Racing Car Simulator) is another:
Udacity - as part of the Self-Driving Car Engineer Nanodegree, also open-sourced their driving simulator (I posted this link elsewhere):
https://github.com/udacity/self-driving-car-sim
I used it in the first term to implement a version of NVidia's End-to-End CNN (https://images.nvidia.com/content/tegra/automotive/images/20...). Something I ran into with the simulator, which may apply to other simulators (though I've played Trigger Rally before with keyboard only and no problems) is that joysticks were really touchy (major fast acceleration on movement of sticks, and no way to really adjust it easily under Linux - which was kinda a disappointment to see - I could do some adjustment, but nothing like what is available on Windows); I found that using a steering wheel controller worked best.