A solo project which I created a fork off of Unity’s machine learning solution, ml-agents. Using this base I created a baseball training camp in which the machine will learn the optimal way to hit the ball
Generation Optimization
To increase the amount of generations that could be ran in an hour I refactored my original code to calculate the distance the ball would fly instead of Unity’s physics engine spending the time making the ball fly.
Anaconda Workspace
Unity’s ML-Agents api is ran through python and is converted into actions in the game. For this I set up a conda workspace which will allow me to keep seperate projects under the same ML-Agents fork without having to reinstall the packages again.
Leave a Reply