Machine learning (sort of!)

My first experiment with AI in games was an air hockey game I published sometime in 2018. It was fairly rudimentary – the AI would just chase the puck and try to hit it. There was no real “intelligence” behind it.

In this new version (download link), it’s a bit smarter…

In the new “Learning Edition” of Air Hockey, the Blue puck is still “dumb” with the same basic “follow the puck and hit it” logic as before, and its movement speed & update rate (how often it senses the puck) don’t change.

However, the Red puck is now a learning/”smart” AI – it uses the same “follow and hit” logic as the Blue puck, but will alter its speed and update rate based on its performance.

Will the “smart” Red puck be victorious after an hour of play, or will the dumb Blue puck foil it through luck and physics?

If you want to know even more, there’s lots of detail on the Red puck’s “smart” logic in this text file

Leave a Reply

Your email address will not be published. Required fields are marked *