I've been working on a game in my free time for a little while now, and I've become really interested in artificial intelligence in games. Programming the A* algorithm for pathfinding was the most fun I've ever had programming. Anyway, I'm at the point in my college career where I'm thinking about what I want my honors thesis topic to be. I'm really interested in team-based AI, but I'm not sure what kind of original work I could do in that area, and I need to narrow it down much more than that. I guess I'm just looking for someone to help me brainstorm, to spark some idea in my mind.

ok heres some words then hope it help.
emergence, genetic, fuzzy, fsm, nueral, sleepy, spatial partitioning, food, hirarchy, so tired.
also dont know if this has been done or if its what ur after but i just pulled it of the top of my head trying to use as many BIG words as i could...
“Coordination of autonomous agents in highly dynamic and interactive environments”

I'll make one assertion. The point of the final year project is to get good grades.

I made the mistake of choosing something really cool to do in AI, but then found that the pressures of the other parts of the course in the final year meant I did not actually do as well as I should have.

It's far better to pick something you are familar with and run with it if you can - or at least something you know you can pick up very quickly.

It's also very important to pick an advisor you get on well with, so if the tutor for that subject is known for being harsh and perhaps in previous years students with projects with them didn't do so well - that's another warning sign to steer clear.

If you know you can do the team based AI then thats great. If you can do a project applying A* - something you have experience and fun with - in an interesting way then you might even be able to use that as a basis.

Perhaps something like "Team search and rescue: How team members using a combination of the A* algorithm and inter-team communication might find their objective quicker then one avatar on its own".

Something like that would mean you can start off with something simple - a big empty space with one avatar stuck in the middle (victim) and your team coming in on one point in the map, agreeing orders and keeping touch on their radios to check for covered ground.

To begin with you could do it without even proper visualation - could be just a big grid and then getting the team to generate a plan. Then counting how many steps it took to find the thing to be extracted.

Extensions would include obsticals that cannot be seen at time of planning, varying the numbers of the team, getting the objective location wrong so they have to set their own new objectives to find the thing, etc, etc.

I'm assuming that you have a professor who is supposed to supervise your project. Find a professor you like, and find out what their research focuses on. Ask them if they have any small projects that would be suitable for an honors project. They have a much better idea of what is achievable in the time you've got than you do; Students ALWAYS over-estimate how much they can get done on their honors projects. Also, your prof will have a good idea of whether a project is worth doing, i.e. can get some decent results out of it. Since it's related to their research and their own interests they will be in a better position to offer advice, and they're more likely to want to help as well. Finally, if you want to do graduate studies, you'll have a useful reference and you might even be able to expand on the work in your project.

The above advice is all sound, so no need for me to reiterate it. I will add though, that it is important to understand what an Honours year research project is all about; it's an introduction to research methodology. What is most important is that you learn and apply basic research skills during this year. That means you need to do the following:
1) Pose a research question;
2) Review current literature relevant to this question;
3) Propose and implement a method of answering the question;
4) Test, using your implementation, the quality of the answer;
5) Present your results in a research thesis.

Now, at Honours level, this does not have to be new contribution to the given field. It can be an investigation of a previously developed technique or application and indeed, these make the best projects at this level, because they allow you to tackle the aforementioned methodology around a structured problem and solution.

Talk to your potential supervisors (if you haven't chosen one already) and discuss with them the sorts of problems they think would be appropriate for you, given your background undergraduate studies.

Cheers,

Timkin

Thanks, you guys have been pretty helpful. I've already talked to a few professors that I might want as my thesis advisor, and I've got a bit of a background in research already (I've been working at my school's research institute since the beginning of this summer). I feel I have a solid footing, I'm mostly just looking to narrow my topic. I'm going to talk to the AI professor tomorrow about it, and I wanted to go in with a variety of ideas.