On 01/22/2013 11:21 PM, Azamat Shakhimardanov wrote:
> Hi Nick,
>
> * JntArray has a public member "data" which is of type Eigen::VectorXd, you can view JntArray as an enhanced Eigen::VectorXd with some additional operations.
I see, it seems I was looking at old API
> * One of such operations is "MultiplyJacobian", which takes Jacobian (which has member "data" of type Eigen::Matrix) and returns Twist. I assume that is what you need, or?
no, this is not what I need, I don't need a twist, but just again a jntArray, look at it as scaling the jointarray, not projecting it to a cartesian space
> * Concerning other versions I don't know their purpose either. Most (if not all) KDL algorithms take plain JntArray types, even if the inputs are joint rates and
> accelerations.

>
> Hope this clarifies things a little.
> Azamat
>
> ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------
> *From:* Dominick Vanthienen <dominick [dot] vanthienen [..] ...>
> *Sent:* Tuesday, January 22, 2013 5:44 PM
> *Subject:* [Orocos-users] KDL JntArray
>
> Hi,
>
> I've some questions (and hidden suggestions) about KDL JntArray:
> *There seems to be no possibility to multiply a jntarray with a matrix, that results in again a jntarray (use-case: control: ydot=K(yd-y) ), is this a deliberate choice?
> *There seems also not to be a 'data' option, as in the other KDL types: you can't get the underlying Eigen type straight away, is this a deliberate choice?
> (especially the combination of above is less practical)
> *There is also a JntArrayVel and JntArrayAcc: does anyone use this? can all algorithms on JntArrays use these types?
> I didn't do any recent testing, but I remembered that there were problems with it
>
> thanks for your replies!
>
> Nick
>
> --
> Orocos-Users mailing list
> Orocos-Users [..] ... <mailto:Orocos-Users [..] ...>
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-users
>
>