[Rock-dev] Fwd: Re: Robot visualization in Vizkit3D

[Malte Wirkus]

Thank you all for your responses.

Back to my initial question: If you are interested in what I provided. 
What I read from the responses, the majority says 'no', because there 
are better file formats available or other libraries that might be used 
for rendering.

Jakob's point is interesting:

 > I did a hack back then for mars models, which is in simulation/mars_viz.

So, something akin is already done. Then of cause it's not necessary to 
use my software for ROCK. One could rather think about integrating URDF 
support into the existing software.
Unfortunately I can't find it. That is maybe because I don't really know 
where to look for it (I looked at 
http://www.rock-robotics.org/master/pkg/index.html and used the search 
functionality of the ROCK website).
As a ROCK newbie, I think it should be easier to find and use, even if 
it's not perfectly aligned with the theories on how such a software 
would look like in the optimal case. The reason for this is that I think 
a robot visualization is, a quite essential _tool_ for people working 
with robots.. as with most tools, it's better to have one than not 
having one.

Best,

Malte



On 05/07/2013 06:23 PM, Jakob Schwendner wrote:
> Yep, we had a discussion on this a while back. I needed exactly what you
> are proposing now:
> A multibody robot model with forward kinematics calculation based on
> joint angles.
> At that time we agreed with the Mars people that this would be a good
> interface for starting to use the same visualisation between mars and
> vizkit.
> I did a hack back then for mars models, which is in simulation/mars_viz.
> In any case, I think we discussed several ways of how we could introduce
> this in rock:
> (a) Vizkit plugin, that takes URDF/Collada/whatever file as a parameter,
> and joint angles as input type. Least amount of effort, but only useful
> for visualization.
> (b) A more tightly integrated approach, that takes the transformer into
> account. I think our general way of configuring and setting up robots
> would benefit from this, too.
> this might benefit from further discussion and could be a good
> opportunity to get different parts working together.
> cheers,
> Jakob
>
> On May 7, 2013 at 4:44 PM Matthias Goldhoorn
> <matthias.goldhoorn at dfki.de> wrote:
>  > Collada was several times an good candidate not only for the
>  > visualization also for mars (also the physics/sim part).
>  > Since noone has time to to this this get forgotten.
>  > If you have some free recources please grap the discussion on
>  >
>  > http://rock.opendfki.de/wiki/WikiStart/OngoingWork/RobotDescriptionFile
>  >
>  > and reply to the ML. Malte L. (ex. R.) was invited to because there was
>  > some open questions for/with mars but i not remember them yet.
>  >
>  > If there _are_ free recources we should definelty define an wider used
>  > and accepted format. There was several limitations on URDF (see link
>  > above).
>  >
>  > Matthias
>  >
>  > On 07.05.2013 16:38, Sylvain Joyeux wrote:
>  > > -------- Original Message --------
>  > > Subject: Re: Robot visualization in Vizkit3D
>  > > Date: Tue, 07 May 2013 15:47:33 +0200
>  > > From: Sylvain Joyeux<sylvain.joyeux at dfki.de>
>  > > Organization: DFKI RIC
>  > > To: Malte Wirkus<malte.wirkus at dfki.de>
>  > >
>  > > On 05/07/2013 09:56 AM, Malte Wirkus wrote:
>  > >> for a software project I am working on, I am developing a
> visualization
>  > >> of multi-body robots based on Open Scene Graph (OSG).
>  > >> Since I did not find something similiar in ROCK's Vizkit3D, I
> wonder if
>  > >> you might be interested if I make this available as a Plug-In for
> Vizkit?
>  > >>
>  > >> Here are some maybe relevant things about it:
>  > >>
>  > >> * As robot description format it used URDF
>  > >> (http://www.ros.org/wiki/urdf)
>  > >> * For kinematic calculation, KDL is used
>  > >> (http://www.orocos.org/wiki/orocos/kdl-wiki)
>  > >> * Therefor there are dependecies on OSG, URDF, a URDF parser
> library, a
>  > >> URDF to KDL parser library
>  > >> * All these libraries but OSG are available in ROS
>  > >> * It makes use of the ressource_retriever from ROS
>  > >> (http://ros.org/wiki/resource_retriever), but can also be used
>  > >> without. Absolute paths to the 3D meshes have to be set in the URDF
>  > >> file then.
>  > >>
>  > >> Forward kinematics are calculated within the visualization.
> Practically
>  > >> that requires joint data (current positions) as input for the plug
> in.
>  > >> Having it as part of ROCK one could also say that forward
> kinematics are
>  > >> calculated elsewhere, and the input for the plugin would be 3D
> poses for
>  > >> each link of the robot. This is the way they do it in ROS (with RVIZ,
>  > >> http://www.ros.org/wiki/rviz).
>  > >>
>  > >> Regards,
>  > >>
>  > >> Malte
>  > >
>  >
>  >
>