[ODE] Car Simulation(Lots of problems)

[Alen Ladavac]

>Yeah, I think this is the friction model. You can actually get
>the wheels to stand still, and the car to coast forward just a
>smidgen per frame. I haven't dived into detail on this -- it
>COULD be something stupid, like me using the wrong fDir1 and
>Slip2 parameters, but I think it's something subtler.

I'm asking because I was tuning some cars for our game, and we use a similar
model (of you are using the findex part in ODE), and we also have an
iterative solver, and I never noticed such an effect. If you are using
findex (and not that part with fixed force), this should not happen.

>Nope, just steering. I did add the different steering angles
>for inner/outer wheel after I built that demo, though, and
>that improves things a little bit.

You really mean different steering, or do you mean different driving speeds
(differential gear effect)?

>The main source file is in the archive, so minus points for
>asking ;-) It's dWorldQuickStep, with 0.01 time step.

Eh, but it is easier to ask than to look. And it is a trait for a programmer
to be constructively lazy. ;) Anyway, I thought that quickstep should not
exhibit problems like that (see above). Did you check it agains direct
solver? Perhaps it is just me tweaking more parameters in our code, so now
it looks better. Then again, perhaps some parameters could be tweaked in
ODE's quickstep as well. ;)

Alen