Printable View
I ran across this article, and was curious what more experienced programmers thought of it?
STRIPS: A New Approach to the Application of Theorem Proving to Problem Solving..—.. AiGameDev.com
New to me.
One reviewer said:
STRIPS is a good theory, but not so good in a practical application.
Is that the general consensus?
Well symbolic ai is cumbersome and slow in general. Unless you have a very simple model this kind of planning is probably too slow for most realtime games, but fine for turn based. The main problem is that whenever an action is taken the model changes so you have to rebuild the plan, or assume that everything is going to end up fine.
Its basically just another search algo, but with pre-conditions, so if you know minimax or a* you should be able to pick it up fairly quickly. If you want to find out more related info google "block worlds" or "micro worlds" I havent actually implemented STRIPS myself yet, just read about it.
It's bloated buzzware. Anyone that has ever actually worked with commercial AI could tell you it will never perform acceptably in a limited resource environment like games.
Any serious paper begins with an immediate in depth description of the technical differences. That link read more like a spam advertisement for natural male enhancement.
Recent ... WTF !?! This guy is actually claiming he invented planning techniques RECENTLY????Quote:
The recent incursion of planning techniques into the world of game development can be attributed in part to the ideas behind STRIPS.
Planning techniques have been around for thousands of years.
That is the concensus - STRIPS is not so good for anything that's time constrained.
Thanks, guys! :)