AI LEARNS BASKETBALL

Поділитися
Вставка
  • Опубліковано 1 чер 2024
  • An A.I. learns how to throw basketballs. The A.I. was programmed using Unity3D (a game engine) with Machine Learning (Reinforcement Learning) using the ML-Agents framework. The Algorithm used to train this A.I. is called PPO (developed by OpenAI). If you are interested in more information, check out:
    / tf-jam-with-ml-agents .
    My name is Sebastian Schuchmann from Germany and I hope you enjoyed this video.
    If you want to support me! :)
    Patreon: www.patreon.com/user?u=25285137
    Twitter: / sebastianschuc7
    Github: github.com/Sebastian-Schuchma...
    If you want to get in contact with me, go to my Webform :)
    www.sebastian-schuchmann.com/...
    Music: GoodNite X Oshi - Homecoming
  • Наука та технологія

КОМЕНТАРІ • 22

  • @revimfadli4666
    @revimfadli4666 4 роки тому +3

    Dude Perfect better watch out...
    Btw, would you upgrade it to play the full basketball game, with teams & rules?

  • @patmw
    @patmw 3 роки тому

    I love your videos man keep it up :)

  • @hermanlindqvist7238
    @hermanlindqvist7238 4 роки тому +1

    I concur this was great, danke!

  • @PickentCode
    @PickentCode 4 роки тому

    This is awesome! Can't believe you only have 3 subscribers.

    • @SebastianSchuchmannAI
      @SebastianSchuchmannAI  4 роки тому +1

      Hey thanks! Really motivating. I will try to make another video soon!

    • @abcqer555
      @abcqer555 4 роки тому +1

      Make that 9!

    • @abcqer555
      @abcqer555 4 роки тому +1

      @@SebastianSchuchmannAI Definitely. Great content, keep it coming :D

  • @simonstrandgaard5503
    @simonstrandgaard5503 4 роки тому

    Beautiful.

  • @tee54985
    @tee54985 3 роки тому

    how did you create the reward. isn't the time lag between throw and hit bad for training? or do you calculate that immediately when throwing?

  • @parikshitsaikia5355
    @parikshitsaikia5355 4 роки тому

    hey did you used reinforcement learning?, btw that was awesome.

    • @parikshitsaikia5355
      @parikshitsaikia5355 4 роки тому

      and if we consider that the player can move anywhere on the ground, what are the action, observation spaces?

    • @sps014
      @sps014 4 роки тому +1

      Looks like NEAT

    • @sps014
      @sps014 4 роки тому +1

      @@parikshitsaikia5355it also may be QLearning

    • @sps014
      @sps014 4 роки тому +1

      Action Space:Vec3 force
      Observable : Distance Vec3 between player and Rod ,May be current Velocity of player
      I am guessing these things, i may be wrong,
      waise yahi hona chahiye shaayad.

    • @revimfadli4666
      @revimfadli4666 4 роки тому

      @@sps014 Proximal Policy Optimization, it's in the description

  • @JuXuS1
    @JuXuS1 4 роки тому

    veri gut

  • @SempiternalScientist
    @SempiternalScientist 4 роки тому +1

    The AI didn't learn basketball, it just learned how to navigate a program you created that simulates a basketball game...there's a major difference.