[ODE] Programmer request (car physics)

[Pancini, Ettore]

if i use exagerate values with dMassTranslate i get
pendolum effect (sort of) so something is actually 
changing. nonenthless my car still flips so i'll try 
the geom transform advice. 



-----Original Message-----
From: Jon Watte [mailto:hplus-ode at mindcontrol.org]
Sent: giovedì 29 luglio 2004 10:05
To: Pancini, Ettore; ode at q12.org
Subject: RE: [ODE] Programmer request (car physics)



I don't think dMassTranslate actually is paid attention to 
by the body functions. At least, I've never made it work. I 
just move the body (and then move the geom back, using a 
Geom transform).

Yes, you need to set fdir, as mentioned in the documentation; 
you can actually set it either in the rolling or anti-rolling 
direction.

I don't include angular velocity in my rolling slip, because 
a wheel in contact with the surface will have a linear 
velocity that's proportional to the angular velocity and the 
radius. If you want to include both, at least include two 
separate scale factors so you can tune them. I only use the 
linear velocity of the body.

The documentation suggests that slip should be linear with 
velocity, not quadratic. I don't know what the actual physics 
is.

I do not enable slip1 -- the way the slip is explained (and 
the way the documentation suggests using it), slip is specific 
to the direction crosswise from rolling direction. Slip1 is 
the direction you're rolling in.

Cheers,

			/ h+




-----Original Message-----
From: Pancini, Ettore [mailto:epancini at ferrari.it]
Sent: Wednesday, July 28, 2004 12:28 AM
To: ode at q12.org
Cc: Jon Watte
Subject: RE: [ODE] Programmer request (car physics)



- I use dMassTranslate() to move the center of mass. am I wrong?
- in using sideways slip I would mention to set the proper friction
directions,
  that is: set fdir1 as the direction the tyre is rolling in... correct?
- In Slip2 = 0.004 * v, v is velocity? do you think is ok doing
something like:
  v = linearVelocity.length() + angularVelocity.length() ?
- Is it correct to consider Slip2 linear with velocity. (I mean, it
could be 
  Slip2 = k*v*v or something). what do you think?
- What values is best for Slip1? shall we set it the same as Slip2?

thanks



-----Original Message-----
From: ode-bounces at q12.org [mailto:ode-bounces at q12.org]On Behalf Of Jon
Watte
Sent: martedi 27 luglio 2004 08:32
To: Alex Amsel; ode at q12.org
Subject: RE: [ODE] Programmer request (car physics)



Can't help on a paid basis, but here's what I've found:

1) You should lower the center of mass of the main body. This is done by
lowering the entire body, and raising the collision geom (maybe using a
GeomTransform). Lower it so the center is just above the ground for that
ground-hugging feel. Make sure it's still POSSIBLE to flip the car,
though :-)

2) You want sideways slip, something like Slip2 = 0.004 * v, in your
contacts. If not, then you'll easily roll the car.

3) You want the Approx contact friction model, because the other model
just limits the force exerted, rather than calculating something based
on actual contact force.

That being said, the behavior of the kind of car I've been able to make
with a Hinge2 and a sphere for each wheel, and a box for the body, is
still pretty beach-buggy-like, rather than race car; even on flat
ground. I've had better success simulating cars with a box and four rays
in the past.

Cheers,

			/ h+


-----Original Message-----
From: ode-bounces at q12.org [mailto:ode-bounces at q12.org]On Behalf Of Alex
Amsel
Sent: Monday, July 26, 2004 5:41 PM
To: ode at q12.org
Subject: [ODE] Programmer request (car physics)


Hey all, been a web-based reader on ODE due to a small internal 
project here so firstly thanks for the various helpful peeps posting.

As have others I've noticed, we've run into a few teething troubles 
with race car physics that are driving one of our programmers around 
the bend (excuse the pun).

I'm looking for someone prepared to help out over the next few days - 
primarily just getting some parameters right but possibly more & 
longer term bit-part work if we decide to build things up.

Work would be paid for so if you already have such things running in 
a stable manner and you can help us stabilise the system then give me 
a shout.

Email me off-list (alex at tunatech dot com)

Regards



Alex Amsel 
Tuna Technologies Ltd (Sheffield, UK)
PC/Console Game Development
Tel: +44 (0)114 266 2211  Mob: +44(0)7771 524 632

_______________________________________________
ODE mailing list
ODE at q12.org
http://q12.org/mailman/listinfo/ode


_______________________________________________
ODE mailing list
ODE at q12.org
http://q12.org/mailman/listinfo/ode