Here is an outline description of a very ambitious project. This is far too big for an undergraduate or MSc project, but it may be possible to select a small subset for such a project. If you have a relevant idea you can try to persuade me to supervise you.
Moreover, I am unwilling to supervise projects based on programming languages on which I am not an expert, since otherwise I will not be able to give good advice and help with debugging will take far more time and effort.
That means I expect students supervised by me to use Pop11, which I know very well or perhaps Prolog which I don't know so well, or maybe Lisp, which I know a little.
In particular I don't want to supervise projects based on languages not designed specifically to support AI, as I think they make the tasks too difficult, and constrain what can be achieved.
The Poplog system which contains several AI languages and the SimAgent toolkit is now available free of charge from here, including a version for PCs running linux: http://www.cs.bham.ac.uk/research/poplog/freepoplog.html
and my recent research papers can be found here http://www.cs.bham.ac.uk/research/cogaff/
And recent research presentations, at conferences, etc. here: http://www.cs.bham.ac.uk/research/cogaff/talks/
If you browse these files you'll get some idea of my expertise and interests. There are also many suggestions for possible topics listed in online TEACH files, accessible from VED, e.g. TEACH PROJECTS (though many of those would be too elementary for a final year or MSc project.)
I do not wish to supervise projects involving neural nets, since I don't know enough about them to give good advice on technical details, and I am probably not a good person to supervise projects involving evolutionary computation. In any case, such projects may require more computer power than we can easily provide.