Flytosky
Convertible environment for reinforcement learning with Kerbal Space Program
Install / Use
/learn @under-control/FlytoskyREADME
Kerbal Space Program - complex environment for Reinforcement Learning
Below you can find medium article with whole story and short tutorial how to start.
Files description:
log - tensorflow session with results csv
a3c_continous.py - a3c network with workers
config.py - setting parameters and connections
ksp_env.py - game environment
requirements.txt - Python 3.6
results_plotter.py - plotter updated continuously
tracker.py - shows the data in console and inside the game
Piotr Kubica - Machine Learning Engineer at Whiteaster
