Re: clock stepping and ODE

ODE uses a first-order integrator, and first-order response constraints. 
This means that time step variability will introduce energy in joints 
and contacts.

To solve the seeming discrepancy between real-time and stability, use a 
quantized main loop.

http://www.mindcontrol.org/~hplus/graphics/game_loop.html

Cheers,

          / h+


Kevin Quigley wrote:
> On Nov 11, 2007 4:42 PM, Jon Watte (ODE) <[email protected]> wrote:
>   
>> 3) To make a simulation stable, make sure that you're using a fixed time
>> step. Variable time step sizes, say based on the clock, will introduce
>> energy into the system.
>>     
>
> This was from the "sliding box on plane" thread earlier this week, but
> I was wondering what kind of instability using the system clock would
> generate?  The user's guide specifically mentions that "ODE is
> designed to be used in interactive or real-time simulation."  That
> implies that ODE will be able to use an outside clock in some fashion,
> although it shouldn't be variable stepping if you want to keep it
> real-time.
>
> Did I misunderstand something from the previous post?  I was hoping to
> use ODE on an RT system.
>
> quig
>
>
>