altugi at March 19th, 2007 19:50 — #1
Is there a preferrable method in AI programming for the behavior of opponents in coaching sims?
For example, how does a manager in the Football Manager series decide what tactic to use in a match, when to substitute a player, what players to shortlist or when to approach to sign a player? Is there a dominant method, or is it a combination of many methods in AI programming?
dark_storm at March 20th, 2007 14:00 — #2
You can use any number of methods. There isn't a preferrable one persay.
Depends on what you are familiar with and how smart you want your manager AI to be.
For instance, you could make it a rule based machine, where if something happens, it directly triggers a response. You could use a finite state machine. If you wanted to be clever and make it hard for the player, use a genetic algorithm or neural networks that can learn. or a combination of any of the above. Its really just what you are comfortable with and what you know how to use.
altugi at March 26th, 2007 06:02 — #3
If you wanted to be clever and make it hard for the player, use a genetic algorithm or neural networks that can learn.
Thanks for the reply. I guess for what I have in mind I should consider genetic algorithms and/or neural networks. I want the rival managers to adjust to the players style in time. Basically they should be able to fetch the data of certain events, analyse them and select those that brought in better results.