More on Ethercat and Orocos

Hi,

just for your information we have built a mobile platform here at IOC. We have
use Orocos (2.5.x), xenomai (2.5.x) and SOEM.

The project (master thesis) to make the mobile platform moves was done by a
very good student. Here is the memo of the thesis (english) [1]. I will try to
publish the code soon.

However, I have to say that we had not be able to make SOEM work with rtnet.
And it's a pending task.

I will inform more about that and publish some code.

Just for curiosity, some weeks ago someone offers some money to clean this
components and take out all the ros dependencies. Finally, the job was done?

Regards,

Leo

[1]https://sir.upc.edu/wikis/roblab/index.php/Laboratory/TheBMMControllingThePlatform

More on Ethercat and Orocos

On Fri, 31 May 2013, Leopold Palomo-Avellaneda wrote:

> Hi,
>
>
> just for your information we have built a mobile platform here at IOC. We have
> use Orocos (2.5.x), xenomai (2.5.x) and SOEM.
>
> The project (master thesis) to make the mobile platform moves was done by a
> very good student. Here is the memo of the thesis (english) [1]. I will try to
> publish the code soon.
>
> However, I have to say that we had not be able to make SOEM work with rtnet.
> And it's a pending task.
>
> I will inform more about that and publish some code.

Interesting piece of information! Thanks :-)

I will discuss this work with the "youBot people" I am connected to,
because this seems too good of an opportunity to miss out on some very good
cooperation!

A technical question: was the normal Linux Ethernet stack not good enough
(especially in the probably only point-to-point network connection
situation that is common in control busses)? My experience is that mobile platforms
typically have such slow natural dynamics that one can easily do without
real hard realtime support.

> Just for curiosity, some weeks ago someone offers some money to clean this
> components and take out all the ros dependencies. Finally, the job was done?
>
> Regards,
>
> Leo
>
> [1]https://sir.upc.edu/wikis/roblab/index.php/Laboratory/TheBMMControllingThePlatform

Herman

More on Ethercat and Orocos

A Dissabte, 1 de juny de 2013, Herman Bruyninckx va escriure:
> On Fri, 31 May 2013, Leopold Palomo-Avellaneda wrote:
>
> > Hi,
> >
> >
> > just for your information we have built a mobile platform here at IOC. We
have
> > use Orocos (2.5.x), xenomai (2.5.x) and SOEM.
> >
> > The project (master thesis) to make the mobile platform moves was done by
a
> > very good student. Here is the memo of the thesis (english) [1]. I will
try to
> > publish the code soon.
> >
> > However, I have to say that we had not be able to make SOEM work with
rtnet.
> > And it's a pending task.
> >
> > I will inform more about that and publish some code.
>
> Interesting piece of information! Thanks :-)

It was a pity that when you were here you didn't see the platform. But, well
it's normal, short visit for a ph. dissertation, meeting with the tribunal and
that's all, time pass so quickly !!!

> I will discuss this work with the "youBot people" I am connected to,
> because this seems too good of an opportunity to miss out on some very good
> cooperation!

Ok, perfect.

> A technical question: was the normal Linux Ethernet stack not good enough
> (especially in the probably only point-to-point network connection
> situation that is common in control busses)? My experience is that mobile
platforms
> typically have such slow natural dynamics that one can easily do without
> real hard realtime support.

we have used the _normal_ linux ethernet stack. However, I would prefer a rt
stack, I think that some vibrations problems that we have could come from
there.

Regards,

Leo

More on Ethercat and Orocos

Hi,

> > However, I have to say that we had not be able to make SOEM work with rtnet.
> > And it's a pending task.

What are your problems with rtnet? We are using SOEM on Xenomai 2.6.2.1 together with rtnet to control an industrial robot via CoE.

We are only using the Xenomai native skin (no Posix skin), so we changed all socket, bind, ... calls in SOEM to rt_socket, rt_bind, ..., and that more or less was it.

Recently, SOEM 1.3.0 was released which offers better support for realtime. In this version, some critical syscalls concerning time measurement have been extracted so they can be more easily substituted by appropriate Xenomai calls.

Michael

More on Ethercat and Orocos

A Dissabte, 1 de juny de 2013, Michael Vistein va escriure:
> Hi,
>
> > > However, I have to say that we had not be able to make SOEM work with
rtnet.
> > > And it's a pending task.
>
> What are your problems with rtnet? We are using SOEM on Xenomai 2.6.2.1
together with rtnet to control an industrial robot via CoE.

Well, simply our platform moved strange and the communications with the servos
didn't works well. I have to admit that could be many things, and I'm
investigating it.

> We are only using the Xenomai native skin (no Posix skin), so we changed all
socket, bind, ... calls in SOEM to rt_socket, rt_bind, ..., and that more or
less was it.
>
> Recently, SOEM 1.3.0 was released which offers better support for realtime.
In this version, some critical syscalls concerning time measurement have been
extracted so they can be more easily substituted by appropriate Xenomai calls.

Ok, do you have soem 1.3.0 patched to run against rtnet? Please, could you
provided the patched files?

Regards,

Leo

More on Ethercat and Orocos

Hi,

> Ok, do you have soem 1.3.0 patched to run against rtnet? Please, could you
> provided the patched files?

I have attached three patch files for SOEM 1.3.0, all need tob e applied in SOEM main directory with -p1 option.

Patch1 patches a bug in the generic Linux implementation. We had this bug in combination with a youBot, after a while (when the buffer was full) it was no longer possible to receive mailbox messages.

Patch2 contains our changes for rtnet. In order to apply patch2, you need to copy oshw/linux to oshw/xenomai and then apply the patch.

Patch3 is a dirty hack to make the youBot work again with SOEM 1.3.0. Apparently the youBot does not completely adhere to the EhterCAT standard and sends mailbox replies which are interpreted as error messages. This patch simply ignores any EtherCAT mailbox error. This patch is not necessary for the other patches to work.

Regards,

Michael

More on Ethercat and Orocos

A Dilluns, 3 de juny de 2013, Michael Vistein va escriure:
> Hi,
>
> > Ok, do you have soem 1.3.0 patched to run against rtnet? Please, could you
> > provided the patched files?
>
> I have attached three patch files for SOEM 1.3.0, all need tob e applied in
SOEM main directory with -p1 option.
>
> Patch1 patches a bug in the generic Linux implementation. We had this bug in
combination with a youBot, after a while (when the buffer was full) it was no
longer possible to receive mailbox messages.
>
> Patch2 contains our changes for rtnet. In order to apply patch2, you need to
copy oshw/linux to oshw/xenomai and then apply the patch.
>
> Patch3 is a dirty hack to make the youBot work again with SOEM 1.3.0.
Apparently the youBot does not completely adhere to the EhterCAT standard and
sends mailbox replies which are interpreted as error messages. This patch
simply ignores any EtherCAT mailbox error. This patch is not necessary for the
other patches to work.

Thanks a lot,

I will apply it to my head git version. I have a CMakelized version of SOEM,
so I will incorporate all the changes. When I finished it, I will send the
link. However, this changes should be incorporated to upstream ...

Regards,

Leopold

More on Ethercat and Orocos

Hi Leo,

That was me. It turned out to be too hard a task for what I was
offering (Reuben tried). Peter quoted it as 24 hours of professional
consulting time at a reasonable rate for hiring a pro to work on
tedious stuff (I'm not sure if the quote is public info), but in the
end I was able to get it to build in our system with some quick and
ugly hacks (like building it with ROS on another machine, manually
ripping out the ROS related stuff, and adding those files to version
control). We're deprecating the one piece of hardware that uses soem
anyway, and I made fast enough progress once I allowed myself to use
"worst practices" on that one component. BTW, building soem with
clang has some bug related to single instance classes in factories
pattern, so I use g++ for building soem (all of our other code builds
with clang).

Cheers,
Andrew

On Fri, May 31, 2013 at 4:38 PM, Leopold Palomo-Avellaneda
<leopold [dot] palomo [..] ...> wrote:
> Hi,
>
>
> just for your information we have built a mobile platform here at IOC. We have
> use Orocos (2.5.x), xenomai (2.5.x) and SOEM.
>
> The project (master thesis) to make the mobile platform moves was done by a
> very good student. Here is the memo of the thesis (english) [1]. I will try to
> publish the code soon.
>
> However, I have to say that we had not be able to make SOEM work with rtnet.
> And it's a pending task.
>
> I will inform more about that and publish some code.
>
> Just for curiosity, some weeks ago someone offers some money to clean this
> components and take out all the ros dependencies. Finally, the job was done?
>
>
> Regards,
>
> Leo
>
> [1]https://sir.upc.edu/wikis/roblab/index.php/Laboratory/TheBMMControllingThePlatform
>
>
> --
> --
> Leopold Palomo-Avellaneda <leopold [dot] palomo [..] ...>
> Institut d'Organització i Control de Sistemes Industrials -IOC-
> Universitat Politècnica de Catalunya -UPC-
>
> Institute of Industrial and Control Engineering
> Technical University of Catalonia
> Avda. Diagonal 647, pl. 11
> 08028 BARCELONA (Spain)
>
> Tel. +34-934017163
> Fax. +34-934016605
>
> --
> Orocos-Users mailing list
> Orocos-Users [..] ...
> http://lists.mech.kuleuven.be/mailman/listinfo/orocos-users
>

More on Ethercat and Orocos

A Dissabte, 1 de juny de 2013, Andrew Wagner va escriure:
> Hi Leo,
>
> That was me. It turned out to be too hard a task for what I was
> offering (Reuben tried). Peter quoted it as 24 hours of professional
> consulting time at a reasonable rate for hiring a pro to work on
> tedious stuff (I'm not sure if the quote is public info)

It was not public, and I think that it's better. OTOH all of us take benefit
of all the work that have done a lot of people for free, so I think that it
was reasonable that someone that has bills to pay made a quote for something
so specific.

> but in the
> end I was able to get it to build in our system with some quick and
> ugly hacks (like building it with ROS on another machine, manually
> ripping out the ROS related stuff, and adding those files to version
> control).

:-(

> We're deprecating the one piece of hardware that uses soem
> anyway, and I made fast enough progress once I allowed myself to use
> "worst practices" on that one component. BTW, building soem with
> clang has some bug related to single instance classes in factories
> pattern, so I use g++ for building soem (all of our other code builds
> with clang).

that's interesting, could so explain what problems with clang?

Regards,

Leo

>
> On Fri, May 31, 2013 at 4:38 PM, Leopold Palomo-Avellaneda
> <leopold [dot] palomo [..] ...> wrote:
> > Hi,
> >
> >
> > just for your information we have built a mobile platform here at IOC. We
have
> > use Orocos (2.5.x), xenomai (2.5.x) and SOEM.
> >
> > The project (master thesis) to make the mobile platform moves was done by
a
> > very good student. Here is the memo of the thesis (english) [1]. I will
try to
> > publish the code soon.
> >
> > However, I have to say that we had not be able to make SOEM work with
rtnet.
> > And it's a pending task.
> >
> > I will inform more about that and publish some code.
> >
> > Just for curiosity, some weeks ago someone offers some money to clean this
> > components and take out all the ros dependencies. Finally, the job was
done?
> >
> >
> > Regards,
> >
> > Leo
> >
> >
[1]https://sir.upc.edu/wikis/roblab/index.php/Laboratory/TheBMMControllingThePlatform
> >
> >
> > --
> > --
> > Leopold Palomo-Avellaneda <leopold [dot] palomo [..] ...>
> > Institut d'Organització i Control de Sistemes Industrials -IOC-
> > Universitat Politècnica de Catalunya -UPC-
> >
> > Institute of Industrial and Control Engineering
> > Technical University of Catalonia
> > Avda. Diagonal 647, pl. 11
> > 08028 BARCELONA (Spain)
> >
> > Tel. +34-934017163
> > Fax. +34-934016605
> >
> > --
> > Orocos-Users mailing list
> > Orocos-Users [..] ...
> > http://lists.mech.kuleuven.be/mailman/listinfo/orocos-users
> >
>