I am creating an android game out of a partner based trick taking card game. What are some generic approaches to AI players for these types of games and what are the advantages/disadvantages of each?
\$\begingroup\$ \$\endgroup\$
3 
- 1\$\begingroup\$ possible duplicate gamedev.stackexchange.com/questions/17904/… \$\endgroup\$jhocking– jhocking2012-02-13 21:36:07 +00:00Commented Feb 13, 2012 at 21:36
- 1\$\begingroup\$ Is the partner also an AI instance, or a human? It's important because the strategy you use in such a game depends on your assumptions about your partner's behaviour. e.g. in bridge, certain play sequences convey information to your partner - good play requires you to handle this communication \$\endgroup\$mikera– mikera2012-06-13 01:42:44 +00:00Commented Jun 13, 2012 at 1:42
- 1\$\begingroup\$ For the initial version your partner will also be an AI instance if the game does well enough or our client just wants to we will update with multiplayer options. \$\endgroup\$lathomas64– lathomas642012-07-04 01:24:28 +00:00Commented Jul 4, 2012 at 1:24
Add a comment |
1 Answer
\$\begingroup\$ \$\endgroup\$
Implementing an AI for a trick taking game would pretty much just involve implementing the same strategy a human player would use. In this you're in luck, as trick taking game strategy has been widely covered, and I'm sure you'd be able to find a wealth of information on strategies for games like Bridge and 500.
You're helped even more by an AI's ability for perfect memory of cards played, and ability to accurately calculate card probabilities. The hard part would be coding in the subjective cues that partners use to convey their actions, such as leading a low card with the expectation of your partner trumping it.
Your best bet in my opinion would be to find a strategy which is heavy on maths and probability measuring, and use that as the basis for your AI.
