This was a very challenging approach and I was so happy to see RL trained AI agents in the game… The moment you mentioned RL I knew it was gonna be tricky… Great job on achieving this much consistency in such a short amount of time… Would love to chat more about it… I infact used all these RL stuff PPO, D4PG etc during my MTech thesis…. +100 for the effort and idea… Great job man… :)
Viewing post in Deep Down jam comments
Nice too see that you liked the game !
Yeah it was pretty tough too get a "feeling" that you can control the characters. In fact, at the beginning, they were responding to their roles, but didn't care about following the leader or the flag !
Yeah sure, would love to exchange about all this with you :) You can find me on the AI & Games Discord with the name Doriens.