Simple unity ml project where an agent (cube) learns to jump
It was my first ML unity project. I followed an online guide in order to understand how the toolkit works. Each cube has a behaviour and detects the obstacle which it is supposed to jump over, as well a "wall" above, the road and a "wall" connected to the obstacle (so we can give him a reward for jumping over)
To run the learning process i create an environment through anaconda.
Run simulations
mlagents-learn (first time)
mlagents-learn --force (restart from scratch)
mlagents-learn --resume (continue the learning process)
The rates were not good tho.
