Re: [Eighthfleet-development] Algorithm Apologies and Thoughts

[Nathan C. <n_...@ho...>]

No need to be so apologetic. =3D)

Angular movement is still wonky. I'll have another go at that tonight =
sometime. There were two major problems in the linear versions that I =
think also could have translated to the angular versions, since I =
translated the derivation straight across. The distance p was supposed =
to be the distance from the current position to the midpoint, which I =
treated wrongly at first, and the lack of the special case examining =
velocity when the position was between p and the goal produced some =
erroneous behavior. I tried to use the same logic for the fixes applied =
to the linear code in the angular code but it didn't seem to work as =
expected before my last commit.

One optimization I can think of off the top of my head is to =
pre-calculate acceleration available in every direction, updated when =
engines are added, removed, or stop working for whatever reason.

One thing I will need is a way to thrust at less than 100% of engine =
power in a given direction. This is necessary to use during the timestep =
when p is less than a full thrust away. I was going to write it but if =
you want to go ahead. Something that takes a simple 0-1 percentage =
should do the trick.

Do you want to work on higher-level nav behavior? For instance, when a =
ship enters the MIN_DIST from its target it shouldn't stop decelerating =
if it wants to actually stop at that target, and we need a way to switch =
off from NAV_GOTO to NAV_STAY when the ship is comfortable with its =
position. Another thing is that if a ship is traversing several points =
it may not want to slow down and try to stop at each point. It may want =
to slow down somewhat according to a provided constant, or depending on =
where it thinks its next goal point is relative to the current point. =
There's definitely a lot of work to do at that level in tickAI and navXX =
functions.
  ----- Original Message -----=20
  From: TIMOTHY K HAMILTON=20
  To: eig...@li...=20
  Sent: Sunday, November 26, 2006 9:39 AM
  Subject: [Eighthfleet-development] Algorithm Apologies and Thoughts


  Alright, so it would seem that my contribution to navigation =
algorithms is to break them, sorry about that. =20

  Your method for lineral acceleration is very obviously superior to =
mine, the movement once it reaches the target point is much smoother and =
oscilates less.  Angular movement still acts a little funny though, as =
it doesn't start decelerating until it's completely passed the target =
angle, instead of approx. halfway there.  I'm not going to touch that, =
however.

  What I will do is try to optimize things a little better.  I'll keep =
my hands out of the angular velocity code, but I think I can streamline =
the linear velocity stuff.  One way I think to keep from doing a lot of =
re-calculating is just to pre-calculate as much of the information as =
possible during initialization of the ship.  One of the most obvious =
values for this, I think, are the various availableThrusts, which only =
need to be re-computed whenever an engine is added or destroyed. =20

  Again, I'll be a lot more careful about touching navigation, and just =
focus on what I promised about tactical behavior instead. =20

  Sorry I broke your code.

  -Tim



-------------------------------------------------------------------------=
-----


  =
-------------------------------------------------------------------------=

  Take Surveys. Earn Cash. Influence the Future of IT
  Join SourceForge.net's Techsay panel and you'll get the chance to =
share your
  opinions on IT & business topics through brief surveys - and earn cash
  =
http://www.techsay.com/default.php?page=3Djoin.php&p=3Dsourceforge&CID=3D=
DEVDEV


-------------------------------------------------------------------------=
-----


  _______________________________________________
  Eighthfleet-development mailing list
  Eig...@li...
  https://lists.sourceforge.net/lists/listinfo/eighthfleet-development