♝ ♞ What's the most advanced A.I. acheived with GameSalad to date? ♞♝
StormyStudio
United KingdomMember Posts: 3,989
Evening world..
Little question about the best A.I. in GameSalad...
This is something I've been thinking whilst planning out the basics of a new 'obviously original and exciting' game idea. There's a simple game mechanic at heart though to be at all note worthy I think the in game characters A.I. will need to be pretty well done.. I sense a lot of prototyping and messing with tables to build lots of characteristic variables that don't unbalance gameplay.
To date the only A.I. I've ever implemented in a game is if you walk within a certain distance of an enemy he will try and attack you... get far enough away and he will stop... otherwise he will just walk back and forth.
Defining advanced A.I. in a Game seems a bit tricky as really any A.I. we design is made more advanced by simply adding more variables that trigger different actions, not exactly self aware.
*The below A.I. levels aren't what I require for my own game but make it easier to talk about.. just giving a nice little break down.
A.I. level 1:
Automated character walks back and forth on a given path
A.I. level 2:
Add rules so they know to walk around moving obstacles and through doors to reach a position.
A.I. level 3:
Add rules so they know to hide in the shadows when a player is near.
A.I. level 4:
Add rules so they work as a team with other in game characters, blocking the exits.
A.I. level 5:
Go back in time to kill John Connor.
So out of interest in seeing what has been achieved so far:
What's the most advanced Artificial Intelligence (A.I.) acheived with GameSalad in a game to date?
Please share either your own games A.I. achievements, other GS games you've played or ideas on the subject.
Little question about the best A.I. in GameSalad...
This is something I've been thinking whilst planning out the basics of a new 'obviously original and exciting' game idea. There's a simple game mechanic at heart though to be at all note worthy I think the in game characters A.I. will need to be pretty well done.. I sense a lot of prototyping and messing with tables to build lots of characteristic variables that don't unbalance gameplay.
To date the only A.I. I've ever implemented in a game is if you walk within a certain distance of an enemy he will try and attack you... get far enough away and he will stop... otherwise he will just walk back and forth.
Defining advanced A.I. in a Game seems a bit tricky as really any A.I. we design is made more advanced by simply adding more variables that trigger different actions, not exactly self aware.
*The below A.I. levels aren't what I require for my own game but make it easier to talk about.. just giving a nice little break down.
A.I. level 1:
Automated character walks back and forth on a given path
A.I. level 2:
Add rules so they know to walk around moving obstacles and through doors to reach a position.
A.I. level 3:
Add rules so they know to hide in the shadows when a player is near.
A.I. level 4:
Add rules so they work as a team with other in game characters, blocking the exits.
A.I. level 5:
Go back in time to kill John Connor.
So out of interest in seeing what has been achieved so far:
What's the most advanced Artificial Intelligence (A.I.) acheived with GameSalad in a game to date?
Please share either your own games A.I. achievements, other GS games you've played or ideas on the subject.
Comments
Guru Video Channel | Lost Oasis Games | FRYING BACON STUDIOS
Guru Video Channel | Lost Oasis Games | FRYING BACON STUDIOS
I like your point about it being the human element that is hard to program, not the simplicity of a computer opponent following a set routine. I guess a set routine with purposefully built in errors becomes more human.
I think your right about most of the games recently have very little to no A.I. .
Perhaps as individual and teams get more used to using tables and attempting bigger games, it might start getting put into games.
I also believe tables are giving GS almost endless possibilities. I'm discovering that to my uses now!
There have been several users who have done an excellent job. @FryingBaconStudios already said that he created a system for Puck It, and I have to say, it's amazing. So Congrats to him.
My GameSalad Academy Courses! ◦ Check out my quality templates! ◦ Add me on Skype: braydon_sfx
Guru Video Channel | Lost Oasis Games | FRYING BACON STUDIOS
http://jamie-cross.net/posts/ ✮ Udemy: Introduction to Mobile Games Development ✮ Learn Mobile Game Development in One Day Using Gamesalad ✮ My Patreon Page
I think I may be a bit tied up if I end up trying to create some 'The Sims' type characters that get on with their day whilst a player interacts with things.
Not A.I. related but I lost a lot of hours to Qwak on the acorn electron...
After a quick google image search it looks like it got remade for the Amiga and then again for the iOS app Store...
What I've been thinking about doing is setting up a table system/automation system for more realistic AI movement and environment interaction. I've thought of several ways of achieving this.
One would be to spawn (randomized or not) actors that talk to a table and fill it with info for the npcs on a per scene basis (so lets say 5 actors that can be reused to auto update cells with x/y/other variables needed), and the NPCS reference that table to get a position of interest or other values by doing a ran(1,5) every once in a while to choose a table row to call from. Essentially the table is just a scratch table in this scenario.
The other one would be a table with pre-defined values for these points of interest that contains all of them for the entire game. That way each scene would have a col value (scene 1= col1, scene 2=col 2, etc) and can reuse a similar random function to choose the values to pull.
If using the first method, magnitude could also play a factor and send things like an override to stop a follow movement and re-start when whatever it is following is too far away. This would be good for an intelligent party of characters, so one is the active, and then however many AI's following. But if they get close enough to the points they go on there own and stop at the pre-defined spot and start activating logic triggered by that position. When you start to move again they go right back to following etc.
I can see benefits and downfalls to both, but I think when it comes time to work on that system i'll go with the scratch table method to save time. Maybe by then i'll have an even better solution.
I've started working on a test project to see if I can create a way to allow me to vary the A.I. from game to game, so different characteristics can be mixed up on different actos so each time you play you get a unique setup. ... it makes sense for the game I'm looking to make.
Think of it this way. Do you remember Ian Livingstone and Steve Jacksons books?
Choose your own adventure? You roll a dice and make decisions - choose a path - some other books were called pick a path.
What you do is the same with your actors - give them paths to take - they have a multitude of choices. Selected by you based on some attributes.
The more steps, choices made, the more organic it becomes.
It sure would be fun to play with
I'm early 31 or 32 ... can't actually be 100% as I stopped paying much attention after I was 21.
Progress on my non linear story/character setup is evolving... trying to design it so once its up and running I can easily add another element to it with out breaking what already existed.. getting there, have'nt had to think this much since doing the 3d pseudo blocks...
I want to build a set of creatures that have different types of movement (straight line, bounding in leaps towards the player, zig zagging towards player, etc), and different sorts of attacks (melee swipe, pounce attack, ranged attack), as well as some neutral behaviours (wander in a small area, look around), and perhaps retreat as well.
I haven't started yet, because it's a bit daunting! Wish me luck.
Here's my game so far...
I'm 35, and I remember the really basic AI of the earliest games. But it's always been about perception. If you can use the right finesse of timing and reactions, they can seem human, and apply the right sort of tension on the player to keep them engaged and in suspense.
I remember being really impressed with the different player characters, and enemy AI's in Commandos: Behind Enemy Lines. Great level design and mechanics in that game.
Commandos ... another awesome game. I guess it got super seeded by 'Command and Conquer' a few years later.