ocl for Stäubli

Dear people,

I think that was Ruben Smits did it, but i would like to ask to the list. In
the ocl directory there's a /hardware/staubli component. I have some
curiosity in this component.

This component use serves to move a Stäubli RX130 robot, RX series. We had two
RX90 and I remember that they had a Adept controller with a t of 16ms. You
could send positions of the joints to the controller using a serial interface
(as the component I think) to move the robot with the alter program, but we
got some problems because a limitation of the serial interface and the data
to send.

Please could you explain more about this component and how it works?

regards,

Leo

Ruben Smits's picture

ocl for Stäubli

On Thursday June 5 2008 18:12:44 Leopold Palomo-Avellaneda wrote:
> Dear people,
>
> I think that was Ruben Smits did it, but i would like to ask to the list.
> In the ocl directory there's a /hardware/staubli component. I have some
> curiosity in this component.
>
> This component use serves to move a Stäubli RX130 robot, RX series. We had
> two RX90 and I remember that they had a Adept controller with a t of 16ms.
> You could send positions of the joints to the controller using a serial
> interface (as the component I think) to move the robot with the alter
> program, but we got some problems because a limitation of the serial
> interface and the data to send.

We "hacked" the Adept controller, meaning:

tapped the joint encoder positions, bypassed the drive-signals (input for the
highpower motor drives) and some I/O for the brakes, powerenable etc.

these hacked signals are connected to ni-DAQ cards and used through comedi in
the /hardware/staubli component.

We only use the serial communication to get the absolute position of the robot
(the result of the adept "where" command), at the configuration of the
component.

This way we get an interface to the robot with a much higher performance.

> Please could you explain more about this component and how it works?

The component is designed the same way as the kuka components, it has a
command prepareForUse waiting for the user to release the emergency buttons,
some methods to unlock/lock, start/stop the axes control and ports to write
the jointpositions and read the desired axes drive values.

Ruben

Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm

ocl for Stäubli

A Divendres 06 Juny 2008, Ruben Smits va escriure:
> On Thursday June 5 2008 18:12:44 Leopold Palomo-Avellaneda wrote:
> > Dear people,
> >
> > I think that was Ruben Smits did it, but i would like to ask to the list.
> > In the ocl directory there's a /hardware/staubli component. I have some
> > curiosity in this component.
> >
> > This component use serves to move a Stäubli RX130 robot, RX series. We
> > had two RX90 and I remember that they had a Adept controller with a t of
> > 16ms. You could send positions of the joints to the controller using a
> > serial interface (as the component I think) to move the robot with the
> > alter program, but we got some problems because a limitation of the
> > serial interface and the data to send.

First of all, thanks for the answer and sorry for the delay . I read it.

> We "hacked" the Adept controller, meaning:
>
> tapped the joint encoder positions, bypassed the drive-signals (input for
> the highpower motor drives) and some I/O for the brakes, powerenable etc.
>
> these hacked signals are connected to ni-DAQ cards and used through comedi
> in the /hardware/staubli component.

so, as I can understand, you have opened the Stäubli controller (a big box)
and inside the box, you have removed the cables in the way that you have the
connexions from the "controller computer" to the HighPower, the decoder from
the robot to the controller, and you have connected this to a DAQ from NI,
connected to a computer. Right?

Then this card is "drive" by a comedi driver, using xenomai or rtai or linux.
Right?

> We only use the serial communication to get the absolute position of the
> robot (the result of the adept "where" command), at the configuration of
> the component.

the the adept computer "also" receive the signals from the robot and also is
working, no? But, if you received the signals from the robot, you "know"
where is the robot. You don't need the "where".

> This way we get an interface to the robot with a much higher performance.
>
> > Please could you explain more about this component and how it works?
>
> The component is designed the same way as the kuka components, it has a
> command prepareForUse waiting for the user to release the emergency
> buttons, some methods to unlock/lock, start/stop the axes control and ports
> to write the jointpositions and read the desired axes drive values.

But, the control? do you have the dynamic model of the rx? or do you have a pd
for each drive? do you send torques to the drives?

The kdl doesn't have dynamic, I think ..

Regards,

Leo