Re:To win at Starcraft, machines need to figure out how to lie
by mort96 on Saturday April 23, 2016 @06:32PM
(#11557709)
It would literally need just one frame to see and store all available information in said frame. This means that in just a few frames, it could jump around the entire map and get a complete overview of what's available. It could easily be abstracted away, meaning the AI has access to a complete view of the map from its point of view, provided by a relatively simple conventional piece of software which just now and then jumps across the entire map and processes the images. It's thus not really an interesting problem to solve in this context, just a minor annoyance.
[ Parent
| on HN ]
There are no comments on this comment yet.