How much more do we need to beg to hear some news for the AIpack
by Exones · in General Discussion · 07/04/2006 (11:28 am) · 8 replies
How much more do people need to beg to hear some news for the AIpack?
Is it ever going to be released?
Was there a problem with it?
Is it not possible then?
How many more years do we need to wait, cause nobody is getting younger and a few years past already.
Is it so good and advanced, that the creators need so many years?
Thanks
Is it ever going to be released?
Was there a problem with it?
Is it not possible then?
How many more years do we need to wait, cause nobody is getting younger and a few years past already.
Is it so good and advanced, that the creators need so many years?
Thanks
About the author
#2
don't wait for it... work on your own game ai.
07/04/2006 (12:57 pm)
I think it's takes up a lot of time working on it, and the developers have to focus on they're other work first. don't wait for it... work on your own game ai.
#3
Search for articles on the type of ai you need and write it yourself.
-Okashira
07/04/2006 (1:53 pm)
You have to realize how hard it is to make something like an ai pack and how long it would take. I mean there are so many different types of ai it would take forever to cover it all. You would have to do FPS ai, RTS ai, Platformer ai, Sports ai, etc...Search for articles on the type of ai you need and write it yourself.
-Okashira
#4
07/04/2006 (11:27 pm)
I've heard mentions of an AI pack, is their going to be one?? I haven't seen an actuall official GG page on it anywhere. It'd be nice if someone could clear that up for me.
#5
Not released to the general public, just special projects :(
Although last i heard phil was still tweaking the aipack so maybe there is some hope.
07/04/2006 (11:54 pm)
I think it went the way of the Horsepack.Not released to the general public, just special projects :(
Although last i heard phil was still tweaking the aipack so maybe there is some hope.
#6
There is no doubt that the AI pack is the hardest thing ever to make and i totaly respect and realise the amount of work and the difficulties....All i'm asking is to hear something fresh about it.
The thing is that there were supposed to be an AI Pack at some time, and it was supposed to be good too.
Check the TDN link:
tdn.garagegames.com/wiki/TorqueAIPackOverview
and also give a look here:
http://www.garagegames.com/mg/forums/result.thread.php?qt=19187
Yes many people believe about the AI pack, that it went the way of the Horsepack.
If you use the "search" thing of GG and write "AI pack" you'll get some responses from 2004 and some of 06, but nothing special. I think people deserve to know a few things about it.
Is it ever going to be released?
Is it going to work with the Torque Shader Engine?
Can someone of the creators please take responsibility and make an official statement about it.
Can we hear at least something from at least someone?
Thanks
07/05/2006 (2:34 am)
Hello to allThere is no doubt that the AI pack is the hardest thing ever to make and i totaly respect and realise the amount of work and the difficulties....All i'm asking is to hear something fresh about it.
The thing is that there were supposed to be an AI Pack at some time, and it was supposed to be good too.
Check the TDN link:
tdn.garagegames.com/wiki/TorqueAIPackOverview
and also give a look here:
http://www.garagegames.com/mg/forums/result.thread.php?qt=19187
Yes many people believe about the AI pack, that it went the way of the Horsepack.
If you use the "search" thing of GG and write "AI pack" you'll get some responses from 2004 and some of 06, but nothing special. I think people deserve to know a few things about it.
Is it ever going to be released?
Is it going to work with the Torque Shader Engine?
Can someone of the creators please take responsibility and make an official statement about it.
Can we hear at least something from at least someone?
Thanks
#7
But for those waiting for an AI pack, I think you might be better off just writing a simple placeholder AI in the meantime. I wrote one in about 7 hours last weekend. It controls a spaceship. It can either attack, evade, or search. Once it sees another object, it asks itself:
1- am I healthy enough to attack?
2- do I have enough power to attack?
3- do I feel like attacking?
If 2 out of the 3 are "yes", then it attacks. If it's shot, then the threshold for #3 goes down and it feels like attacking less. As it searches, however, it feels like attacking more and more. It also wants to attack more if it successfuly shoots the other object. #2 is set based on how often your power is totally drained. (ie raise the amount of power you require to attack if you keep running low). Etc.. Etc... the thresholds for yes/no to these questions are randomly initialized when the game starts and, as I said, adjusted throughout based on the AI ship's experience.
If the AI ship runs into another ship frequently, then it decides to turn around and try to free itself. This prevents ships from just ramming into each other and sticking there. There's also a low probability random movement that keeps the ship from moving too predictably. It also adjusts the angle-range in which it will fire at another object. That is, if the other object is in range, start fire at X degrees away from the exact angle between the ship and the object. (Though I'm not 100% sure this is working.)
My point is, it only took a day to implement an AI that, while not too bright, is sufficient for testing and can at least get a shot or two into the human player. My only plan at the moment is to make the AI predict where the player is instead of shooting where it's at when it fires (ie don't shoot where the enemy is, shoot where it will be) and make sure it only fires *at* something, not into empty space.
There's also plenty of articles out there on AI. I found this one interesting, partially since I'm studying evolvable hardware and this is a closely related concept: http://www.gamasutra.com/features/20060626/strom_01.shtml
07/05/2006 (1:50 pm)
AI itself isn't hard. "Good" AI on the other hand... ;)But for those waiting for an AI pack, I think you might be better off just writing a simple placeholder AI in the meantime. I wrote one in about 7 hours last weekend. It controls a spaceship. It can either attack, evade, or search. Once it sees another object, it asks itself:
1- am I healthy enough to attack?
2- do I have enough power to attack?
3- do I feel like attacking?
If 2 out of the 3 are "yes", then it attacks. If it's shot, then the threshold for #3 goes down and it feels like attacking less. As it searches, however, it feels like attacking more and more. It also wants to attack more if it successfuly shoots the other object. #2 is set based on how often your power is totally drained. (ie raise the amount of power you require to attack if you keep running low). Etc.. Etc... the thresholds for yes/no to these questions are randomly initialized when the game starts and, as I said, adjusted throughout based on the AI ship's experience.
If the AI ship runs into another ship frequently, then it decides to turn around and try to free itself. This prevents ships from just ramming into each other and sticking there. There's also a low probability random movement that keeps the ship from moving too predictably. It also adjusts the angle-range in which it will fire at another object. That is, if the other object is in range, start fire at X degrees away from the exact angle between the ship and the object. (Though I'm not 100% sure this is working.)
My point is, it only took a day to implement an AI that, while not too bright, is sufficient for testing and can at least get a shot or two into the human player. My only plan at the moment is to make the AI predict where the player is instead of shooting where it's at when it fires (ie don't shoot where the enemy is, shoot where it will be) and make sure it only fires *at* something, not into empty space.
There's also plenty of articles out there on AI. I found this one interesting, partially since I'm studying evolvable hardware and this is a closely related concept: http://www.gamasutra.com/features/20060626/strom_01.shtml
Torque Owner Badguy
good work on the query.