Implementing our Machine Learning AI

Hey monkeys,

we posted details on our machine learning AI and how we implemented it.
Finding resources was pretty hard, so we decided to write this post to help others looking for resources. (We provide many useful sources as well)

This article includes details on how we implement and use:

  • machine learning (a reinforcement learning method called SARSA)
  • hierarchical AI structure (inspired by killzone’s AI)
  • influence maps (for communication with bots & players)
  • a video with our first results beating a simple scripted AI

Here’s the link to our article. It contains many details, so it’s rather long.

(This was previously posted in our blog thread, but it was pointed out that it might interest people who don’t care about our devblog :stuck_out_tongue:)