[ODE] bending wheels again?

[Nate W]

On Tue, 5 Nov 2002, Nicola Di Leva wrote:

> When it's just a matter of going back and forth, the car behaves quite
> well, but as soon as it starts turning, the wheels progressively lose their 
> alignment with the relative joint axes.
> Previous posts mentioned high rotation speeds and friction coefficients, but 
> this is not the case [...]/

I think you're describing something that I also noticed with the freefall
demo on my machine, and my guess is that it could be corrected by tweaking
the CFM.  Try making it an order of magnitude smaller... e.g. if it's
0.001, change it to 0.0001 (more zeroes).  I didn't actually test this
theory, but it looks similar to what simulations in Juice do when the CFM
is too high, so I think it could be the same thing.

The catch is that turning down the CFM makes the simulation more rigid,
thus less forgiving, thus more prone to instability or explosion.  It's a
compromise.  It may well be possible to stiffen it a bit before any
problems appear, it's impossible to say without trying.  

-- 

Nate Waddoups
Redmond WA USA
http://www.natew.com