Submitted a simple version of Q-learning under branch Hang.
In this version, all states are discrete, haven't take continuous location under consideration, it is only a Pacman in the 3D version with multi-agents now. And the parameters still need to be fixed. I will do this after the final exam.
Submitted a simple version of Q-learning under branch Hang.
In this version, all states are discrete, haven't take continuous location under consideration, it is only a Pacman in the 3D version with multi-agents now. And the parameters still need to be fixed. I will do this after the final exam.
The action part was implemented by Jingjing.