Talk:Displacement Vector

From Robowiki
Revision as of 06:30, 16 January 2011 by Jlm0924 (talk | contribs) (→‎How They're Implemented: : Thx Voidious, I'm perfectly clear :))
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Kudos

Thanks for posting this. I thought I understood it before, but now I really get it. Now to figure out a good neural network representation... =) --Darkcanuck 04:49, 24 September 2009 (UTC)

How They're Implemented

I don't think I understand the method of displacement Vectors. It sounds like a bots begining and end position over a given time is translated to be relative to your own bot before recording. Is this what it is? later on you sum the vectors to get a complete displacent vector to aim at? if someone could correct me. thx -Jlm0924 06:48, 17 April 2010 (UTC)

It is the relative position to enemy's robot. Like the enemy has moved for -115.23 pixels in the direction of -23.1 degrees or something like this. To get the aiming angle, just projected enemy's movement vector with the displacement vector. --Nat Pavasant 09:27, 17 April 2010 (UTC)

Right, it's relative to the enemy bot. Also, there is no summing of vectors. To get a firing angle, you just multiply a vector by the number of bullet ticks and apply it to the enemy location/heading. The vector basically measures displacement per tick. --Voidious 14:16, 17 April 2010 (UTC)

Essentially, like a 'guessfactor' except measuring the bot's movement relative to itself, instead of an angle relative to the wave source. --Rednaxela 14:51, 17 April 2010 (UTC)

ahh, cool concept! I'm on the verge of understanding it.. lol. I need a minute to think it throu :) -Jlm0924 03:48, 18 April 2010 (UTC)

The detail I didn't originally realise was you fire waves.. ;) I am curious... When you record a displacement Victor: Did you record the bots entire displacement from the begining to end of a wave, then divide the DV by bullet Time? -Jlm0924 19:28, 15 January 2011 (UTC)

Yep, so the vector I record is the average amount they moved per tick, relative to the initial position. So a bot just going straight at full speed would be (0, 8), for instance. I do use "waves", but the location of the wave source and the abs bearing from wave source to enemy bot don't matter (besides dictating the time interval I record over). Let me know if anything's still not clear, or you could take a look at Diamond's code. --Voidious 21:16, 15 January 2011 (UTC)
Thx Voidious, I'm perfectly clear now :)

Similarity to GF

This really is a great deal like a guessfactor. For example a spinbot has a very predictable movement, for a pattern matcher it is an incredibly trivial pattern, however an unsegmented gf gun or Displacement Vector Gun has more of a problem, for different reasons. For the GF gun it covers an entire span, but spikes at edges of the enemy movement (given more time there), giving them some help. A DV, on the other hand when scaled by the bullet flight time causes the target displaced vector to be outside of the standard range of movement quite often. Averaging many vectors, similar to how bins work produces a better overall match, these averaged displacements equals to about what a GF gun would, and spikes at the edges of the movement pattern. (If its not obvious I made gun that uses these). — Chase-san 13:20, 20 July 2010 (UTC)