Google Research Football: A Novel Reinforcement Learning Environment

  • If I read correctly, the agent only controls one player at a time. On offense it controls the player with the ball, and on defense it controls probably the player closest to the ball. The other players are controlled by the built-in AI. Controlling a single agent kind of takes away from the appeal of deep-RL: that entire teams can learn to coordinate in novel and optimal ways.

  • I guess I don't get it... What does this game have that SC2/Dota doesn't?

    As far as I can tell, the main goal for reinforcement learning is to make it so that it doesn't take 10k learning sessions to learn what a human can learn in a single session, and to make self-training without guiding scenarios feasible.

  • "real bayesians" vs "frequentists united" at 0:33 in the video :D

  • I bet that ai’s will find a lot of physics bugs to exploit early on.

  • Perhaps this will be used in live sports in the future. Giving real time feedback to players for optimum positioning. Would be a cool test but I still prefer to watch sports played the ‘traditional’ way.

  • Wonder if they use the same tech to predict the outcome of football matches. I've seen them show it on the Premier League games.

  • Do all the players have the same skill set?

    Interesting to see how things like faster players change optimal play

  • Any chance for Google Research Rugby?

  • Oh, this is soccer. Madden NFL would be more interesting.

  • > The Football Engine is written in highly optimized C++ code, allowing it to be run on off-the-shelf machines, both with GPU and without GPU-based rendering enabled. This allows it to reach a performance of approximately 25 million steps per day on a single hexa-core machine.

    Missed opportunity to use Rust for memory safety.