Robo Home | Targeting | Changes | Preferences | AllPages

Well quite a few of the methods on the targeting page are out dated and unnessecary, and the only reason I could see keeping them around is as a record of the evolution. From my point of view, we are currently in the age of GuessFactorTargeting, it is by far the most used targeting method amongst the top bots. PatternMatching cannot be discarded, because while it is not used as much, it is proving successful against WaveSurfers and is the best thing for getting incredibly high scores against simpler bots. But there are a variety of other concepts that were more theoretical stepping stones towards techniques like Waves, GuessFactorTargeting, and other modern targeting methods.

When I say that X became Y, I am not saying that this was the sole evolutionary path. Usually several methods converged to make a technique, and even the technique that it evolved to was around at the same time.

LaserTargeting - Originally a less processor intensive version of virtual bullets. At fire, your position and your target's position are recorded, and then the target's position is recorded again at the time the bullet should hit. From that data the appropriate firing angle can be reached. By segmenting based on the angles that are being returned this evolved into GuessFactorTargeting, and by checking the positions every tick instead of only at one point, it became Waves.

BestPSpace - The predecessor to modern GuessFactorTargeting Segmentation, using a variety of seperate maps of variables to decide which distribution yielded the best probability at a certain angle.


Robo Home | Targeting | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited July 1, 2005 15:41 EST by Jokester (diff)