rec.autos.simulators

>I was wondering if anyone else has tried getting the GPL AI to work
>correctly on the ovals.  Wherever ovals in GPL are mentioned on the various
>sites they say it isn't possible due to the AI logic.  I've been playing
>with the *.LP files by hand and I've gotten them to go fast and generally
>smoothed out part of their driving line, but they wiggle pretty bad in
>spots.  Does anyone know of a way to take the data from a replay file to get
>what is needed for the LP files?  It looks like the data is there because
>the GPLDump utility shows a "Lateral" graph that is the same as the graph
>generated from data within the LP file.  I'm not a programmer so I can't
>figure out the data segments in the replay file.  Does anyone know where I
>can find this info?  I know getting them to drive smooth is one thing and
>getting them to pass intelligently is quite another, but ya gotta start
>somewhere.  If someone knows for a fact that getting it to work properly is
>impossible please do let me know as I would like to get back to actually
>driving it instead of teaching it how to drive.  :)

>Thanks!

>-Tom

> Have you been able to get them to work better than those AI files we
> experienced a couple of months ago? The ones where the AI accelerated,
> then crashed into invisible walls, then took off, then crashed, etc.

> Rafe Mc

> On Sun, 12 Mar 2000 02:34:33 GMT, "Tom Bond"

> >I was wondering if anyone else has tried getting the GPL AI to work
> >correctly on the ovals.  Wherever ovals in GPL are mentioned on the
various
> >sites they say it isn't possible due to the AI logic.  I've been playing

>I don't know how they compare to those (I didn't see them), but they do stay
>on the track and generally take the correct line.  The main problem I'm
>having is on the front stretch at C***te (the only one I've messed with
>so far).  As they come out of 4 towards the first dogleg they swerve left
>and right, something like 10-15 feet on the track.  They do sort of okay on
>the frontstretch, then swerve less noticeably through the second dogleg and
>into 1.  They do fairly well through the turns and backstretch, with only
>minor problems.  They usually don't hit the walls (visible or invisible) and
>they accelerate fine.  I'm fairly certain I can correct the problems by
>modifying the data by hand, but it'll take a LONG time.  Oh, and they don't
>pass each other, but that's a different problem I think I can handle once I
>get them driving straight.  I did find a way to use a replay, but it did not
>correct the swerving.

>Tom