🎉 Celebrating 25 Years of GameDev.net! 🎉

Not many can claim 25 years on the Internet! Join us in celebrating this milestone. Learn more about our history, and thank you for being a part of our community!

What do we expect from AI?

Started by
10 comments, last by Magic Card 24 years ago
This is a question I''ve put into my game. In my opinion, it''s counter attacking and dodging. I have played so many games where the enemies just walk int you, and they KNOW you''re going to hit them! It''s like AI with an attack pattern! Instead, why don''t they just stay where they are, dodge your bullets, and lead you on a wild goose chase? And, can''t they hit us? I don''t know how many times I''ve played Jedi Knight and been able to block shots off of AT-ST''s! If they fired faster, it''d be more diffult! What do you think AI needs? Top quality games don''t kick ass as well as these. http://danavision.homestead.com Prepare to be blown away! ------------------------- Magic Card
-----------------------------A world destroyed, a myth rebord. Some truths should remain untold...Check out NightRise today, coming eventually from DanAvision Software Entertainment.http://www.danavisiongames.com
Advertisement
I think AI could be a lot better and in the past years a lot of games have drastically improved their AI. To follow your example several FPS-games do allow the opponents to duck behind crates or use some form of evading your attack instead of bumping into you.

There is however one thing you should be aware of as a game developer. There has to be fun in the game, especially in the single player scenarios.

Take any level of Quake. You start out as a SINGLE player against a level filled with say 40 opponents. If these opponents are only half as good as you then you will not survive the level.

So you have to make the AI 20 times as dumb as the player or put only one opponent in a level that can match up to the player''s potential.

It''s all a matter of balance.

******************************
Stefan Baert

On the day we create intelligence and consciousness, mankind becomes God.
On the day we create intelligence and consciousness, mankind becomes obsolete...
******************************
******************************StrategicAllianceOn the day we create intelligence and consciousness, mankind becomes God.On the day we create intelligence and consciousness, mankind becomes obsolete...******************************
We are working on the AI for a RPG. Thing is : we know how we are going to do it, we just to not know what to do

Here is a short list of things we would like to be able to do :

  • Sub-tactics for fight groups (mainly cops and army)
  • Global tactics for groups (the cops should be able to dispatch patrols and reinforcements in case of a shootout, or to handle a riot, corporate security should be able to organize a search through the building, ...)
  • Communication between NPCs.
  • Complex discussion system (for those who remember Lure of the Temptress''s "ask x to ask y to go to z and bring back the bread to me", well, something similar)
  • Complex movement system (follow x, staying at least 20m behind him, while avoiding cops)
  • Realistic crowd management
  • Breathing life into the background
  • Dynamic NPC generation
  • Dynamic story generation (i.e. random mission generator, except it does not have to be missions as well as newspaper generator) would be nice.



Of course, if you ave any other idea, I''ll be glad to hear them.

Be reading you,
David
Hi all

IMHO RTS 2D like games have more complicate AI then RPG...

This is why:
AI has to:
- build up complicated bases and he has to defend them
- buils large army and then to atack
- has to use very diffrent weapons and tactics
- has to know strategic moves like conquer important bottle necks and expand to important resources while do not alow YOU to do the same....
-has to couter all humman tactics and strategies like :
early rush attack, dodge in and build superior technology,build heavy army and swarm over him, starve him by killing his resources etc

ALL this HAS to be done on maps that YOU Design and have NO predefined paths or strategic info stored into...

You see why AI for RTS (mainly 2D but some 3D also) games
its way more complicated then RPG


Bogdan
obysoft
AI needs better programmers.

-Eck

EckTech Games - Games and Unity Assets I'm working on
Still Flying - My GameDev journal
The Shilwulf Dynasty - Campaign notes for my Rogue Trader RPG

i''ve never played the game (only seen it played for awhile) but in halflife i heard about the ninja type guys further in the game that would do just that - lead you into a better area for them than for you. i hear they were practally impossible in the higher diff. levels.

btw check out the other ai discussions, they may have some revelant information.

-Thr33d
Hmm, I think an RTS AI is simpler than a good RPG AI, since nearly all the stuff in a wargame can be expressed in some sort of points system... evaluate the map to work out which are the best places to capture, assign all units a point value to compare them, or a matrix of X Unit vs. Y Unit effectiveness to determine optimal responses, etc. Whereas an RPG might have to deal with human emotions, natural language, and story generation, as well as some tactical combat considerations, and pathfinding, just as an RTS would need.

It all depends on how far you want to go
Well, I''d say that RTS is hugely simpler to formalize. After all, you know your goal and can express it with little complexity. Then, you can write rules or processing algorithms and experiment.

Doesn''t mean that making a good RTS AI is that simple, though, were it only due to time limitations.

Be reading you,
David
I agree with both Kylotan and Altamann.

Creating a good FPS AI is way much easier than creating a good FPS or RPG one.

I''m investigating (not much at the time) ways to create good AI for RPG games.
It appear that I''ll have to specialise the AI in different fields: talking AI, fighting ones...

Maybe I''m wrong, and I must try to make a human like AI, but I don''t think so.


-* So many things to do, so few time to spend *-
-* So many things to do, so little time to spend. *-
Surely if you find one sort of AI implementation easy it''s because you''re not pushing the envelope.

What you''re saying is that the implementation of the AI at the current level of FPS/RTS/RPG may be more difficult for one than another.
There are so many directions and so much functionality for the future of all of these genre''s that you can make the AI very complex and very difficult, to the point of not-currently-possible. It all depends how much time developers allow for the AI and how much CPU time they''re willing to give up to it.

MikeD

This topic is closed to new replies.

Advertisement