Announcement

Collapse
No announcement yet.

Artificial Intelligence: UniMind

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

    Artificial Intelligence: UniMind

    Hey there, i was wondering that how i can implement a Uni Mind in unreal engine for my A.I System.

    What i mean is i want to implement a system that can decide some of my Allies and Enemies Behaviour.

    So i have this right now :
    1.imagine that we have 10 Enemies, 4 allies
    2. Allies will lock on 2 Enemies with some Sight checks and Distance checks. (Enemies that they can see and closest (1 enemy can be shot by only two ally.)


    Now let's say that i want a unimind to take control of who shot who (including Enemies who are Shooting at me (i dont want all of them shooting me or my allies)





    ((if you have any Knowledge of implementing any kind of unimind in any way, please let me know ❤️
    Last edited by AmiRSlaYeR; 12-22-2020, 12:18 PM.

    #2
    Originally posted by AmiRSlaYeR View Post
    Hey there, i was wondering that how i can implement a Uni Mind in unreal engine for my A.I System.

    What i mean is i want to implement a system that can decide some of my Allies and Enemies Behaviour.

    So i have this right now :
    1.imagine that we have 10 Enemies, 4 allies
    2. Allies will lock on 2 Enemies with some Sight checks and Distance checks. (Enemies that they can see and closest (1 enemy can be shot by only two ally.)


    Now let's say that i want a unimind to take control of who shot who (including Enemies who are Shooting at me (i dont want all of them shooting me or my allies)





    ((if you have any Knowledge of implementing any kind of unimind in any way, please let me know ❤️
    Hi man,
    Its pretty easy , try build something,

    Its a bit strange but a lot of games are already almost "uniminded",
    since using programmation you already can access info from wherever and pass information wherever and take decision based on the whole situation.
    You can build "Puppets" or little A.I, that respond do a list of actions, And use a Mastermind to determine the various action you need for every puppet and assign them.

    Comment


      #3
      Originally posted by Est_engine View Post

      Hi man,
      Its pretty easy , try build something,

      Its a bit strange but a lot of games are already almost "uniminded",
      since using programmation you already can access info from wherever and pass information wherever and take decision based on the whole situation.
      You can build "Puppets" or little A.I, that respond do a list of actions, And use a Mastermind to determine the various action you need for every puppet and assign them.
      Yea Bro i know that way pretty good, i was wondering if there's a better option or solution like using something smarter for ex: an Observer unimind or etc...
      Anyways, Thanks for Answering.

      Comment


        #4
        Originally posted by AmiRSlaYeR View Post

        Yea Bro i know that way pretty good, i was wondering if there's a better option or solution like using something smarter for ex: an Observer unimind or etc...
        Anyways, Thanks for Answering.
        Fantasy-Talking i think that there are at least 2 kind of unimind,
        the overlord: there is one entity that collect info from all the puppet and take decisions. and its pretty constant in cleverness.
        The Cluster: That grow in cleverness by every puppett assimilated ,because every new puppet will host his mind and share calculation , adding

        But create a blueprint to create "The Cluster" is something really hard XD
        i think that the overlord type is pretty good and doable.

        Comment


          #5
          yea I'm actually more interested in Cluster Based Unimind not the overlord, and i'm aware of that overlord is much easier to run(perfomance) and to do i would be appreciated if you do know any references for this articles

          Comment


            #6
            Originally posted by AmiRSlaYeR View Post
            yea I'm actually more interested in Cluster Based Unimind not the overlord, and i'm aware of that overlord is much easier to run(perfomance) and to do i would be appreciated if you do know any references for this articles
            No i dont have any specific reference,
            Not a superexpert here : But you can always fake-one

            Basically you create an overlord-type
            Subdivide all the behaviour is steps and loops and give them a relation with a "structure"
            So you have stats like "" Perception "" Brain "" Strategy ""
            for example you can make a search path toward target with step 1, get the relative direction. step 2-3-4, detect wall and path, 5-10 try pathfinding A , but only with 50 smokerooms, 25-30- use all the smokeroom you need to detect the path.

            Lets say the unimind have 50 units simple like mouse.
            evey mouse have int of 0.1, you have 5 of brain.
            So do the steps until pathfinding but only with 50 smokerooms.

            Same thing for placing the mouses in formation,
            if you have too little brain will only give you 1 location where all mouses will try to gather,
            more brain , will give you 5 , 10 places and divide the mouse in these "formation"
            100 , will use the mouses like a battallion XD

            Hope this help XD

            Comment


              #7
              Originally posted by Est_engine View Post

              No i dont have any specific reference,
              Not a superexpert here : But you can always fake-one

              Basically you create an overlord-type
              Subdivide all the behaviour is steps and loops and give them a relation with a "structure"
              So you have stats like "" Perception "" Brain "" Strategy ""
              for example you can make a search path toward target with step 1, get the relative direction. step 2-3-4, detect wall and path, 5-10 try pathfinding A , but only with 50 smokerooms, 25-30- use all the smokeroom you need to detect the path.

              Lets say the unimind have 50 units simple like mouse.
              evey mouse have int of 0.1, you have 5 of brain.
              So do the steps until pathfinding but only with 50 smokerooms.

              Same thing for placing the mouses in formation,
              if you have too little brain will only give you 1 location where all mouses will try to gather,
              more brain , will give you 5 , 10 places and divide the mouse in these "formation"
              100 , will use the mouses like a battallion XD

              Hope this help XD
              Thanks Brother , Really Appreciate your help man.
              i get that and it's a brilliant way and great example by the way.

              let me say something else now, so let's say we have 1 'Cluster' with 10 members,
              all members will cooperate with each other and decide what to do or what not to do for each member

              now let's say we have 10 'Cluster'
              and all of them are working perfectly individually
              but in some cases some of them needs to cooperate with each other, i mean Cluster To Cluster or something like that.

              so in cases it may be 10 cluster with 10 cooperative members, cooperate with each other to make the best decision

              Comment


                #8
                Originally posted by AmiRSlaYeR View Post

                Thanks Brother , Really Appreciate your help man.
                i get that and it's a brilliant way and great example by the way.

                let me say something else now, so let's say we have 1 'Cluster' with 10 members,
                all members will cooperate with each other and decide what to do or what not to do for each member

                now let's say we have 10 'Cluster'
                and all of them are working perfectly individually
                but in some cases some of them needs to cooperate with each other, i mean Cluster To Cluster or something like that.

                so in cases it may be 10 cluster with 10 cooperative members, cooperate with each other to make the best decision

                I think you could create
                An object " hypothesis ", that is just a whole bunch of variables , like the task , the obstacle of the task , and the resources...
                And one cluster Create an Hypothesis for a task, and than
                Share it with another , to make him too a check.
                Every cluster could check if the part of the hypothesis are better than the one he can provide.
                And adjust the hypothesis , to send it back and again for a number of times.
                Once processed should be the best option the clusterS can came out with .

                This could be really cool but is also creazy hard to make,
                For make a cluster be able to read the data and see that , "The plan is to use the mouse to attack, because they can pass in a little hole in the door"
                --- "But i have snakes that pass too, and are far more deadly, so i propose to use snake instead of mouses.

                Comment

                Working...
                X