Hi all,
We willing to use this to generate some of our velocity profiles :
http://www.reflexxes.com/
As this is quite close to what KDL could to internally I share this here in
case someone would be interested in. I hote it's not a miss use of the list
(I have no link with reflexxes, I just found this recently on the net).
I think the main problem is the licensing.
The library seems quite well designed (espacially in terms of separation of
concerns) from a user point of view (I don't know about the implementation).
Motion Generation Library
On Wed, 26 Sep 2012, Willy Lambert wrote:
> Hi all,
>
> We willing to use this to generate some of our velocity profiles : http://www.reflexxes.com/
> As this is quite close to what KDL could to internally I share this here in case someone would
> be interested in. I hote it's not a miss use of the list (I have no link with reflexxes, I just
> found this recently on the net).
> I think the main problem is the licensing.
That's indeed one problem. The other is that the added complexity is
difficult to justify for all practical cases: the _dynamics_ of the robot
system become so much more important than the kinematics when you want to
optimize smoothness and timing at jerk level... You should start using
dynamics-based MPC methods instead; those are also still missing in KDL,
but we are planning activities on them :-) Others might _have_ something
ready already, in which case I am very interested!
> The library seems quite well designed (espacially in terms of separation of concerns) from a
> user point of view (I don't know about the implementation).
Knowing Torsten Kroeger, you can indeed be sure that it is done neatly :-)
Herman
Motion Generation Library
Kdl does not go that far as reflexxes. No arbitrary initial conditions and no"type 4" motions. On the other hand, I have the impression that refexxes does not do that much for task space motion (besides linear), but I could be mistaken.
In our research we use the Acado toolkit www.acadotoolkit.org to do similar things, but in a more general way, using MPC and variations on MPC.
Best regards,
Erwin
Willy Lambert <lambert [dot] willy [..] ...> wrote:
Hi all,
We willing to use this to generate some of our velocity profiles : http://www.reflexxes.com/
As this is quite close to what KDL could to internally I share this here in case someone would be interested in. I hote it's not a miss use of the list (I have no link with reflexxes, I just found this recently on the net).
I think the main problem is the licensing.
The library seems quite well designed (espacially in terms of separation of concerns) from a user point of view (I don't know about the implementation).