[ODE] Fwd: [ opende-Patches-1336066 ] Joint feedback in quickstep

[Jaroslav Sinecky]

Done. I have just commited my final version of this patch, little bit 
more optimized. It should have now almost zero performance hit if you 
request no joint feedback and use up only as much memory as is needed 
for your feedbacks (in case you use them).

Jaroslav


Jason Perkins wrote:
> FYI...
>
> ---------- Forwarded message ----------
> From: SourceForge.net <noreply at sourceforge.net>
> Date: Nov 10, 2006 9:48 AM
> Subject: [ opende-Patches-1336066 ] Joint feedback in quickstep
>
> Initial Comment:
> In quickstep you don't get correct joint feedbacks in
> case there are more than one joint on a body. For each
> joint the feedback you get is actually sum of feedbacks
> of all connected joints.
>
> This patch computes feedbacks by multiplying the part
> of lambda vector related to the joint in question with
> respective block of J' matrix.
>
> ----------------------------------------------------------------------
>
> Comment By: Remi Ricard (papadoc)
> Date: 2006-11-10 09:48
>
> Most of the patch can be found in the trunk revision 1119.
> The only missing part is the part where only part of the
> Jacobian matrix is copied.
> In the code the complete Jacobian matrix is copied and saved
> to be used later.
>
> ----------------------------------------------------------------------
>
> Comment By: jsinecky (jsinecky)
> Date: 2005-10-24 03:45
>
> I think this patch made it to Unstable (or stable) branch but
> unfortunately in it's premature state. This final version is more
> efficient but AFAIK was not applied to CVS.
>
> ----------------------------------------------------------------------
>
> You can respond by visiting:
> https://sourceforge.net/tracker/?func=detail&atid=382801&aid=1336066&group_id=24884
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
>
>