Re: Turn Based Games

Jason Reposa wrote:
> Soccer sounds good, but a little overwhelming for second year undergrad's.
> Understand that not every student is going to be as excited about AI as you
> are, or nearly as knowledgeable. They will not spend an entire semester
> developing AI for soccer, and then training them. Sure, I'd like them to do
> that. If it was a AI course. This is an introduction. They should spend
> about 20 hours on this assignment. I'm already pushing them hard by saying
> 20. If a student spends 5 hours on their homework it's acceptable, since
> they usually have many other classes to worry about.

It sounds like the soccer simulator might not meet your needs, but I'll 
just throw in one more plug for it.  It's true that your students 
probably won't create very sophisticated soccer teams with 20 hours of 
work, but they can certainly do better than a straw-man "greedy" team 
that just swarms the ball and only takes a few lines of code.  And it 
gives them an opportunity to think about the basic stages in 
knowledge-based decision-making: situation interpretation, goal 
selection/activation, action selection.  I think it's an ideal domain 
for studying simple knowledge-based systems, but probably isn't so good 
if what you are focusing on is weak decision-making methods (although 
there might be some way of encoding soccer states into a heuristic 
function suitable for A*).