Reinforcement learning

Will be adding more details as soon as the report is available.

 

The direction controller and a debug layer that projects the path ahead of the character.

 

 

The direction controller steers the character towards the green dot.

 

 

The direction controller is bound by the motion space provided by the animations, and can therefore only reach a finite set of points. It will try and get as close as possible.

 

Leave a comment