[ODE] Arbitrarily positioning bodies connected by joints

Shamyl Zakariya shamyl at zakariya.net
Mon Jul 30 14:47:39 MST 2007


> In fact, resetting the anchors could
> induce/enhance joint drift since joints are usually not *perfectly*  
> satisfied
> during any timestep..

Not only that, but if you're measuring the "value" of some joint  
( say the compression on a slider which is acting as a spring )  
resetting it on translation could ( and probably would ) change its  
zero point.

I posted a few days ago that there's no trickery required. My  
troubles were coming from my having made a mistake in my  
transformation code, not ODE.

shamyl at zakariya.net
     "Such a theory has to be bizarre and elaborate, as well as being  
stupid"
         -- Jim Loy, regarding a hollow earth


On Jul 30, 2007, at 7:21 AM, Deak Szabolcs wrote:

> "The anchor is specified in world coordinates" - means that *you*  
> specify the
> anchor in world coordinates when you set it. It is then transformed  
> to the
> body space of the bodies connected by the joints, and it is used in  
> the
> relative coordinate form for all internal calculations.
> So both statements are true, and the relevant one in this case is  
> that you
> *don't* have to reset anchors if you transform the connected bodies  
> except for
> those anchors that are set to the world. In fact, resetting the  
> anchors could
> induce/enhance joint drift since joints are usually not *perfectly*  
> satisfied
> during any timestep..
>
> Beau Albiston wrote:
>>  From the ODE manual: "the anchor is specified in world coordinates."
>> If your body frame and world frame are aligned, then your  
>> statement is true.
>>
>> -Beau
>>
>> dan miller wrote:
>>> I do this all the time, without resetting anchors.  Why would  
>>> that be
>>> necessary?  It's relative to the bodies, not global coords  
>>> (unless you're
>>> anchored to the world)
>>>
> _______________________________________________
> ODE mailing list
> ODE at ode.org
> http://ode.org/mailman/listinfo/ode



More information about the ODE mailing list