Quote: Original post by KylotanAnd I seem to remember someone wrote a simple Bayesian predictor for an FPS which, given your current position, very accurately estimates where you're going to go next, meaning it can almost always be waiting for you in an optimal position. It's easy to do.

Sounds intruiging. Got a link or a name?

I can see how that would work. As for researching it, start with getting comfortable with Bayesian networks. The whole idea behind Bayesian stuff is taking observables and combining them with uncertain, yet relatively likely premises, to arrive at what is the most likely outcome.

Once you are familiar with how they work, the implementation should be obvious. At that point, everything becomes problem-specific.

I am familiar with them. I have implemented something similar in RoboCup. I'm interested in this particular implementation because Kylotan mentioned it achieved optimal behaviour, which isn't always easy - at least not given a limited sample size to learn from.

My use of the word optimal wasn't a strict one. After all, who could judge whether the position truly was the best?