[ODE] Basic Questions

[Jon Watte (ODE)]

The impulse given by a force is proportional to force and time, so if
you apply a force F 100 times over 0.01 time steps, you'll get the same
impulse as if you apply F 50 times over 0.02 time steps. Thus, you
should NOT scale the force by the time step.

However, ODE uses a first-order integrator, so it's not time step size
invariant. Resting (but not disabled) bodies keep accelerating into
their supports, and being pushed back by constraints. The amount they
accelerate by varies by time step size, so they will find different
equilibriums at different time steps. Varying time step means they won't
find equilibrium at all.

For examples of things that help with cars (sway bars, lower center of
gravity, slip contacts, ackerman steering, speed-proportional steering,
etc), check out the carworld example:

http://www.mindcontrol.org/~hplus/carworld/

Cheers,

			/ h+


Matthew Harmon wrote:

> 1. Does ODE use the current step time as a factor when applying forces?  
> That is, if I am running at a variable timestep, must I scale my forces 

> 2. I have a vehicle simulation and at high-speeds I get a lot of 
> “bouncing” of the tires.  I think this is caused when a tire penetrates