KDL Screw Theory
Submitted by thoseme on Thu, 2010-07-22 13:11 |
Hello,
I had a look into the KDL Documentation, and at the first glimpse I suppose that calculations are based on Screw Theory. This is because of the uncomplicated manner how to define serial chains, and the data structures for wrenches, twists and so on.
I spent some time in understanding Screw Theory, but I am still not familiar with it, having a lot of questions and being unable to calculate some things I am interested in. In order to get familiar with that, I would like to play around with a mighty tool, like Orocos KDL.
Regarding this, it would be very helpful if there would be some examples. For example, applicating a wrench and looking whether the serial chain will move, or what forces can be seen in which joint.
Did somebody do such calculations, and has done a little bit of documentation?
Best Regards ERik
KDL Screw Theory
On Thu, 22 Jul 2010, linuxrechner [..] ... wrote:
> I had a look into the KDL Documentation, and at the first glimpse I
> suppose that calculations are based on Screw Theory.
If you like to call it that way, yes, it's "screw theory". But that
corresponds exactly to what comes out of a classical dynamics approach, as
it should be. The only difference might be that 'screw theory' considers
all 6D entities as the basic primitives of the theory, while classical
dynamics (for historical reasons only) always makes a difference between
translational and rotational 'vectors'.
> This is because of the uncomplicated manner how to define serial chains,
> and the data structures for wrenches, twists and so on.
>
> I spent some time in understanding Screw Theory, but I am still not familiar with it, having a lot of questions and being unable to calculate some things I am interested in. In order to get familiar with that, I would like to play around with a mighty tool, like Orocos KDL.
>
> Regarding this, it would be very helpful if there would be some examples. For example,
> applicating a wrench and looking whether the serial chain will move, or what forces can be seen in which joint.
The first one requires a dynamic model (its the "inverse dynamics"
functionality that you would be testing by seeing how a chain reacts to a
6D force ("wrench") at the end point; the latter can also be done via
kinematics only properties of the chain, being the "Jacobian transpose"
mapping (which reflects the _static_ equilibrium between the end-effector
wrench and the joint torques required to counteract it, without taking
dynamics into account).
> Did somebody do such calculations, and has done a little bit of documentation?
The documentation is in every robotics textbook; KDL did not invent
anything new in this context :-)
Herman
KDL Screw Theory
Hello,
I had a look into the KDL Documentation, and at the first glimpse I suppose that calculations are based on Screw Theory. This is because of the uncomplicated manner
how to define serial chains, and the data structures for wrenches, twists and so on.
I spent some time in understanding Screw Theory, but I am still not familiar with it, having a lot of questions and being unable to calculate some things I am interested in. In order to get familiar with that, I would like to play around with a mighty tool, like Orocos KDL.
Regarding this, it would be very helpful if there would be some examples. For example,
applicating a wrench and looking whether the serial chain will move, or what forces can be seen in which joint.
Did somebody do such calculations, and has done a little bit of documentation?
Best Regards
ERik