An API to do game AI research in shooters

Question

I'm looking to do a machine learning related course project. I'm basically looking for a framework for a top view 2d shooter game, and apply machine learning algorithms to it.

There is a framework available to do research in car racing, called TORCS, and I was looking something similar to this, but for shooters.

Basically I would like a high level API to make the bot move, shoot, pick weapons etc.

Some of the work that could be done: Lets say you need to model how your bot will fight during combat. You use a neural network to map enemy position, bot's position, bot's ammo, etc to how you should move, and what weapon you should choose.

Is there any (preferable 2D, Python) framework which will help me to do this?

Answer 1

• Robocode in Java or .Net.
• Marvin's Arena in C#, VB, C++
• Brood Wars API, for Starcraft.
• Pogamut 3 GameBots2004 for Unreal Tournament and so.
• Planet Wars / Galcon Clone AI. 2D but no a shooter.