Is there a way to let humans play against agents (trained using self-play) in PettingZoo environments, similar to how the game would be played using the GUI?
For turn-based classical games (like rock-paper-scissors, tic-tac-toe, chess, etc), one can still make it work by mapping specific keys to the actions of the human agent. But how can this be achieved for real-time parallel games, say Atari Pong?
I tried using the idea mentioned above, but waiting for the user's input resulted in a laggy game experience.