"How do you handle "two birds, one stone" types of problem solving using utility AI?" One approach would be to use a scoring function for each of the various desire rating types. Then a high level scoring solution (or data set) that is a set of those scoring functions. You pass all the choices to consider through the scoring solution where each functions result is summed to produce a final score per choice. You pick the highest score and can maybe add some randomness to sometimes pick lower scores just to change things up. Its important that each scoring function always adds positive values rather than ever subtracting. In this way you can mix and match a bunch of desires and the outcomes that are highest will always be the "best" with some tuning.
Alturis posted a topic in Math and PhysicsI was looking to find out if anyone knows of an optimal method to find the squared distance of a path defined by a series of 3D points? Given an arbitrary array of vectors, find the squared sum of the distance between each points and its previous point I mean. To my knowledge in order to do this you have to sum up all the distances and then square the result. But was wondering if there was a tricky way to do this only using the squared distance between each of the points.