Considerable effort in our lab is invested into the task of creating data-driven game AI that can learn from people. There are different ways to achieve this task, and we explore various directions of possible research activities. However, we primarily work with AC SDK, which is the core of Artificial Contender (AC) technology developed by TruSoft. They let us use it for research purposes, and we update and improve AC SDK from time to time. Internally AC SDK is based on a relatively simple representation of game processes with a method similar to a probabilistic finite state machine or a Markov chain. However, AC SDK also includes a large set of tools that indispensable in AI development.
Project repository: ACSDK.