[ODE] nonholonomic vehicle path following
Hendrix_ at gmx.net
Mon Jun 11 10:31:27 MST 2007
You should not copy the steering values from one car to the other
verbatim, but calculate the difference to the desired path and apply
that to the car. This way you will have a self-correcting steering. I
hope this is stable, which depends on how big the errors are from
step to step.
On 11.06.2007, at 08:29, Jon Watte (ODE) wrote:
> Given that ODE is a semi-implicit first-order integrator, much like
> other game physics engines, that is correct. There will always be some
> error. In fact, if the circle is parametrically defined, there will
> always be some error with any kind of numerical integration of this
> kind. The question you need to answer is "how big an error can I
> accept," and then see whether you can tweak the tunable parameters
> (response force, dampening, time step size, etc) to hit that goal.
> That's true for any numerical integration method; not just ODE.
> / h+
> Ke wrote:
>> Thank you for your reply.
>> So our 4-wheeled vehicle simulated by ODE is not able
>> to perform such path following tasks accurate enough.
>> For example, say, we asks the vehicle built by ODE, to
>> follow a given circle path. The trajectory of the
>> vehicle would not be exactly same as the given one,
>> due to the accuracy of ODE's integrator and solver.
>> Please point me out if I am wrong.
>> Thanks again and we appreciate your input.
> ODE mailing list
> ODE at ode.org
More information about the ODE