[ros-users] Fuerte SIG 2.10: Efficient Simulation/Control of Articulated Robots

Wed Sep 14 09:37:25 UTC 2011

On Wed, 14 Sep 2011, Mike Stilman wrote:

>> Indeed. But rather, it is not about "extending", but integrating two (or
>> more) different sets of functionalities.
>
> This may prove to be a difficult challenge indeed. The two libraries have
> significant difference in representation. I would love to discuss it with
> you and Karen together - though I'm really not sure that this whole
> conversation needs to continue to spam the ros-users list - unless folks
> are really that interested.
>
Where should it take place then?

>> Have you ever asked about it? Or made suggestions? That's how open source
>> projects work :-)
>
> I think part of open source is that "how it works" is open to
> interpretation :0}
>
> Honestly, at the time I was looking ~08-09 there were no concrete plans
> to add dynamics to KDL. There was this confusing comment chain:
> http://www.orocos.org/node/735
> which I now find to be even more confusing.

I leave the responsibility for that message to its author :-) And if you
find things confusing, there is no better place to be than in an open
source project, since you can ask questions to resolve your confusion,
knowing that (most of the time) the developers have no "hidden agenda" that
they want to keep from you:-)

> More recently, I checked again on the KDL page and documentation and there
> was no mention of any forward/inverse dynamics. Even now dynamics are
> not mentioned and all the examples are purely kinematic.

That's a very valid remark, since the website documentation is lagging
behind, _and_ our hosting service failed to keep our svn repository up and
running at the "standard" place. But on the mailinglist much information
about dynamics solvers has been circulated. Available now are things like
these:
  <http://bugs.orocos.org/show_bug.cgi?id=666>
  <http://people.mech.kuleuven.be/~rsmits/kdl/api/html/classKDL_1_1ChainIdSolver.html>

> At the same time, working with Karen we have a complete dynamic solver
> with lots of examples (just not releasing until we complete the contact
> handling) - i.e. have a full open source dynamic simulator with both
> dynamic solvers in generalized coordinates and contact resolution. It will
> be accompanied by documentation and fully BSD.
>
>> I react to your posting, and the same thing could/should happen for you
>> with things KDL might or might not provide in the future.
>
> Basically there was intermittent discussion about dynamics in KDL on
> the web and it did not appear that there was a final solution in place
> after a couple years of development. Mostly "experiental." I'm aware of
> those discussions - but I was looking for a tested dynamics solution that
> we could use directly and hence now we have one.

See links above. The code was added in June 2009...

>> If you have unit test cases, and want to share them with others, that
>> would be a nice contribution to the community.
>
> Abslutely we will have lots of test cases to share. The library should
> definitely be entirely open with lots of examples.

Sorry to be picky, but I read lots of "will be"s...

>> These are _integration_ issues, which should not be done by extending one
>> single library. Otherwise, we end up where we have ended up for decades
>> already: huge monolithic libraries that are not interoperable.
>
> I really do think this is a longer discussion. Maybe even one to have in
> person. Will you be at IROS?

No. And I prefer to have discussion on mailinglists, so that everyone can
profit from the discussion, and not just the people that discuss :-)

>> I'm looking forward to these things. Within KDL we are now trying to use
>> Collada the standard data format to represent kinematic chains. (And to
>> couple it to other functionalities such as controllers.)
>
> Yup, same here. Collada and URDF actually to support multi-resolution
> meshes and sensors. Probably starting with Collada first.
>
>>> Exactly, so I would consider a force/impedance controller to be a tool
>>> that may benefit from the dynamics of the library or the collision
>>> checking library or neither or both. I see no reason to force the tools
>>> (algorithms) built on top of the basic solvers to be constrained in any
>>> particular way.
>> Agreed!
>
> Great!
>
>> What does "join the SIG means"? Yet another mailinglist?
>
> At the moment it means go to this wiki:
> http://www.ros.org/wiki/fuerte/Planning
> And add your name to Section 2.10

A wiki is not really a replacement for a mailinglist. The Orocos/KDL
already exists, is widely followed, and on topic for these things. It's up
to you, of course, to decide to use it or not; let me just make a comment
that there is already tremendous fragmentation in open source efforts, and
I fear this one will only contribute to that. So, my concrete suggestion
is: use the ROS or Orocos mailinglist, instead of starting yet another one.

Herman

> ------------------------ Here's a forward of what this means from Sachin
>
> The signup period will last until September 14th.
>
> After the signup period, any SIG that has at least two people signed
> up for it will be considered valid.  If you are proposing a SIG, we
> urge you to recruit authors/maintainers of the relevant software --
> the authority of what goes in/out of a library remains with the
> maintainer/author, so their buy-in is crucial.  Similarly, we
> encourage authors/maintainers to signup for SIGs that are relevant to
> their software.
>
> The follow-up process is also very simple:
>
>  1. Each SIG designates a SIG Coordinator.  If an SIG cannot agree on
> a coordinator, one will be chosen for you.
>  2. The coordinator will organize planning meeting(s) for the SIG.
> This can be over IRC, video chat, at IROS, or whatever medium bests
> fits the composition of your SIG. The deadline for these meetings is
> September 28th.
>  3. Each SIG group will post their planning notes as a sub-page of
> http://ros.org/wiki/fuerte/Planning/<SIG-NAME>.  The deadline for
> posting these notes is September 30th.  SIGs can have followup
> meetings, but this initial deadline is to ensure that the SIG is
> activated.
>
> The plans are just that -- plans.  It is up to the members of each SIG
> to achieve as much of that plan as possible in time for the freeze
> dates of ROS Fuerte (remember, there's always Groovy Galapagos).
>
> As always, we appreciate your participation.  We hope this more
> distributed process will better match the distributed nature of ROS
> development.
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users