Unity ML-Agents 1.0 - A.I. Shooting Game (FULL WALKTHROUGH)

Поділитися
Вставка
  • Опубліковано 6 чер 2020
  • Unity's ML-Agents Machine Learning framework is a great way to get started with the topic. Many people have requested a shooting style game, so here it is.
    Support me on Patreon: www.patreon.com/user?u=25285137
    Keep in touch: / sebastianschuc7
    I have created a Discord Channel for everybody wanting to learn ML-Agents. It's a place where we can help each other out, ask questions, share ideas, and so on. You can join here: / discord
  • Наука та технологія

КОМЕНТАРІ • 57

  • @SebastianSchuchmannAI
    @SebastianSchuchmannAI  3 роки тому +3

    Hello everybody! I have created a Discord Channel for everybody wanting to learn ML-Agents. It's a place where we can help each other out, ask questions, share ideas, and so on. You can join here: discord.gg/wDPWsQT

  • @whoami82431
    @whoami82431 4 роки тому +1

    Your tutorials are really good man! You should consider making videos on other aspects of Machine Learning too (that is, stuff other than Unity ML). Looking forward to more content from you!

  • @jonatan01i
    @jonatan01i 4 роки тому

    That kiwi guy making friends is now everywhere.
    Thank you for the video, it was fun to watch!

  • @DeJMan
    @DeJMan 4 роки тому +1

    Just wanted to say thanks for teaching me MLAgents. I made my own model and upload it to the unity3d subreddit where it's the top post right now. I mentioned you in the comments as well.

  • @kyleme9697
    @kyleme9697 4 роки тому

    Hello Hindsight Sebastian ... LOL !! Great videos, can't wait to see more.

  • @jeremymcdonaldson9896
    @jeremymcdonaldson9896 4 роки тому +2

    Really great video!

  • @mrtnmck
    @mrtnmck 4 роки тому

    soooo good!! Time to revamp good old visual studio 😀Thanks for videos like this one!

  • @alphonsegs
    @alphonsegs 3 роки тому

    Thanks for your videos !! Good Job ! I want to learn more about ML using unity :D !

  • @insohamnia
    @insohamnia 3 роки тому +3

    Hi Sebastian, great content thanks!
    I've one question, when you moved from one level to another (example from static agent/enemy to moving agent enemy, etc.)
    Did you load the old traning model or did you restart training from a new model ?

  • @GaryMcKinnonUFO
    @GaryMcKinnonUFO 3 роки тому

    I love the interjection 'Hindsight Sebastian here' :)

  • @williamdelgiudice9173
    @williamdelgiudice9173 4 роки тому +6

    Great content, you should do a video on training agents in teams like the two vs two soccer demo!

    • @revimfadli4666
      @revimfadli4666 3 роки тому +1

      Yea that seems to be a great sequel to self-play, especially if it's role-based thus you might want several policies, one per role

  • @vaibhavjaiswal799
    @vaibhavjaiswal799 4 роки тому +20

    can you explain ML-Agents from scratch like working of every single line of code for ML agents.....because they are a little bit confusing

    • @samselvaraj8171
      @samselvaraj8171 3 роки тому

      yes please

    • @revimfadli4666
      @revimfadli4666 3 роки тому

      For people new to Unity, new to ML, new to C#, or new to programming?

    • @rickybloss8537
      @rickybloss8537 3 роки тому +1

      @@revimfadli4666 Since this is a ML video I'd say those new to ML should be prioritized

    • @revimfadli4666
      @revimfadli4666 3 роки тому

      @@rickybloss8537 agreed, although that also includes experienced Unity devs, C# devs new to Unity, game devs new to Unity and/or C#, and people new to both

    • @jixs4v
      @jixs4v 3 роки тому +1

      @@revimfadli4666 you shouldn't be experimenting with ml agents without knowing unity and c#

  • @visakhr123
    @visakhr123 4 роки тому

    Hi,
    thank you for this . very useful

  • @codykearse1689
    @codykearse1689 4 роки тому

    Hey Hey, the newest release! Fantastic!

  • @AdnanAhmed-kk2qr
    @AdnanAhmed-kk2qr 3 роки тому

    Great Channel, I am unity3d developer. I am enjoying your work. Keep it Up

    • @umairwaseem882
      @umairwaseem882 12 днів тому

      is there way to generate opponent through ai in run time and make them shoot main player?

  • @visakhr123
    @visakhr123 4 роки тому +1

    Hi Sebastian, We are working on a similar concept and we have used your source code. But when we built the system with two Enemies and one Agent we are facing an issue.
    The Agent is not responding to x translation and z translation. We have actually replicated the code but we're not able to figure it out. But when we tried to open your project and check we found out it is working. So can you please suggest what could be an issue like where we might have gone wrong when replicating?
    Because of the same issue we're not able to train the agent properly as the x translation and z translation not updating the rewards are not great.
    Thank you!!

  • @micabarshap
    @micabarshap 3 роки тому +1

    I have a question: you teach about adding gail to the training_config.yaml config file, but there are more config file in ml agent in Ml unity library ,gail_config.yaml file what that gail file for?

  • @TheAmishWarlord
    @TheAmishWarlord 4 роки тому +12

    Thank you so much for this content! I got a little stuck trying to follow along, is it possible for you to share your code?

  • @WeeADream
    @WeeADream 26 днів тому

    so. is it possible to take what the ml agent learned and add it to ai? or am i stuck with the agent trying to learn?

  • @maxazure
    @maxazure 4 роки тому

    Very impressive.

  • @alvaromachucabrena97
    @alvaromachucabrena97 3 роки тому +1

    Hello !
    I am challenging an FPS video game and I want my enemies to chase me. My enemies would be my agents, but how can I modify your code so that my enemies train to follow me?

  • @NevaranUniverse
    @NevaranUniverse 3 роки тому

    Is it possible to have these 3D raycasts be...well...3D? As in to be able to also detect height, since most games arent exactly only on a flat plane. Perhaps in the latest versions of ML Agents?

  • @pulsarhappy7514
    @pulsarhappy7514 Рік тому

    Hi, I am deligthed to have found your channel and this video as I am trying to do a very similar curiculum learning (for a 2D version of an arena same as yours).
    I am using a camera with a top down view, but unfortunately my model is not improving at all.
    I am using 2 continuous actions for the AI to move its pointer on the screen as X,Y variables, and a last continuous action for the shoot decision.
    I'll keep trying to see if I can find something that works.
    If you ever want to try and make an AI playing the game through a top down camera sensor (or 3rd person for a 3D game), I'd be very interested :) !

  • @nischalada8108
    @nischalada8108 4 роки тому +2

    You should make videos more often :)

  • @MrEmiljose
    @MrEmiljose 4 роки тому

    hey Sebastian, Your video is amazing. I have made this simulation; everything works except one problem. My shooting agent has only rotations motion but no translational motion. But when I use your GitHub repo it's working well. I have used the same code with little tweaks. If you have time, please help me with it.

  • @mrzion4711
    @mrzion4711 3 роки тому

    you should have added a penilty/ reward for ammo used. to help stop the spinning random firing trying to hit a target

  • @mindbodymusicrelaxation9000
    @mindbodymusicrelaxation9000 4 роки тому +1

    Very unique content :D Fully watched & liked. New friend here to stay. Thanks!

  • @jimkim7423
    @jimkim7423 4 роки тому

    good job

  • @elementsofarah7918
    @elementsofarah7918 3 роки тому +1

    how can I see the performance metrics

  • @TheDayDreamer17
    @TheDayDreamer17 3 роки тому

    Hey Sebastian, Can you provide me a suggestion to improve my project, I have created a similar environment where the agent has to rotate, point towards the approaching enemy and also shoot it. But here the agent doesn't stop shooting, I have added a negative reward of 0.03 at initial trials then I increased the value to -0.5, still it shoots continuously. Do you have any suggestions?

    • @koreanjesus2712
      @koreanjesus2712 3 роки тому

      perhaps making the ammo count the same as the enemy count ?

  • @nikhilsharma2236
    @nikhilsharma2236 3 роки тому

    bro please reply, i actually am learning ML, so is it necessary for me to use unity, can't i use unreal???

  • @noncopyrightmusic1570
    @noncopyrightmusic1570 3 роки тому

    got new subscriber

  • @SpriteeoN
    @SpriteeoN 3 роки тому +3

    Watching on 0.25 Speed and still struggling to follow anything haha, great video but would love to see something more like a tutorial :)

  • @Bensmithgardening
    @Bensmithgardening 3 роки тому +1

    AHHH! I cannot see what code you are writing or find it on github so hard to follow!

  • @BramOuwerkerk
    @BramOuwerkerk 4 роки тому

    Really great channel! Could you make an agent which doesn't use raycasts?

    • @adityadas8276
      @adityadas8276 4 роки тому

      Just instantiate an object(bullet) in front of the gun and add some force...the physics does behave weirdly though...wouldnt recommend it

  • @nl03
    @nl03 3 роки тому

    Is it possible to download this?

  • @The_Ministry_of_War_Network

    ML agents are team vs team?

  • @pytchoun140
    @pytchoun140 3 роки тому

    How to have the source code ?

  • @ps5games821
    @ps5games821 4 роки тому

    WOW Top videos make more plese
    Great

  • @babycake10
    @babycake10 3 роки тому +1

    Video is too slow, need to bit faster.

  • @rafaelportugal5261
    @rafaelportugal5261 3 роки тому +1

    "Beware of the scribes, which desire to walk in long robes, and love greetings in the markets, and the highest seats in the synagogues, and the chief rooms at feasts;"
    "Which devour widows' houses, and for a shew make long prayers: the same shall receive greater damnation." Luke 20:46-47
    "For he is not a God of the dead, but of the living: for all live unto him."Luke 20:38
    Repent ye: "For the wages of sin is death; but the gift of God is eternal life through Jesus Christ our Lord." Romans 6:23
    Jesus Christ loves you: "For God so loved the world, that he gave his only begotten Son, that whosoever believeth in him should not perish, but have everlasting life." John 3:16
    "And he said unto them, Take heed, and beware of covetousness: for a man's life consisteth not in the abundance of the things which he possesseth."Luke 12:15
    "For there is one God, and one mediator between God and men, the man Christ Jesus;"I Timothy 2:5