This week, we looked into Markov Chains which use matrices to create incremental changes to base values. This is mainly used within economic statistics but it’s also really handy for creating AI. If you combine Finite State Machines (FSM – we’ve been learning these for three years now) and the Markov chains method, we can create individual and interesting AI.
The workshop we did where we used Pokemon as an example of Markov chains was slightly confusing as I haven’t played a lot of Pokemon and so wasn’t sure what everyone was going on about in most cases. I understand the actual chains principle so at least that’s something…
This session was useful as it helped to recap previous lectures we’d looked at and help to move this into applicable situations within Games Design. I feel relatively comfortable that I could design something like today’s lecture for an AI character. Depending on what direction my project goes in, I may try and prototype some FSM and implement the Markov chains method to show how my AI would react. I’d like to involve some innovative AI in whatever I do this semester as these sorts of NPC actions could help separate your game from someone else’s.
For notes on the lesson see Lectures/Crits/Talks book, light pink tag labelled Markov Chains, pages 35 – 41.