[ros-users] Weirdness in URDF tutorials

John Hsu johnhsu at willowgarage.com
Wed Dec 29 10:50:53 UTC 2010


I remember that robot_state_publisher does not publish any transforms unless
there is at least a revolute or prismatic joint in your urdf... Wim, can you
confirm this?
John

2010/12/29 Björn Giesler <bjoern at giesler.de>

> Hi,
>
> well frankly, I think the documentation is very good but it makes quite a
> few assumptions. I've worked through the tf tutorials, and while they're
> great, I'm missing a step-for-step tutorial for simple platforms that don't
> have laser scanners. The tutorial for the navigation stack seems great, but
> it makes many many assumptions about system components that just aren't
> there if you don't have a laser scanner. Once I've worked through all of
> this, I may write a non-laserscanner tutorial.
>
> I'm still confused though how the whole tf system works. Right now, I've
> built a simple URDF file for my robot that contains the base link and one
> foot, here it is:
>
> <?xml version="1.0" ?>
> <robot name="r2d2" xmlns:xacro="
> http://playerstage.sourceforge.net/gazebo/xmlschema/#interface">
>  <link name="base_link">
>    <!-- robot coord origin is between the main feet -->
>    <!-- just put a tiny box there -->
>    <visual>
>      <geometry>
>        <box size=".01 .01 .01"/>
>      </geometry>
>    </visual>
>  </link>
>  <link name="left_foot">
>    <visual>
>      <geometry>
>        <box size=".36 .175 .16"/>
>       </geometry>
>      <material name="white">
>        <color rgba="1 1 1 1"/>
>      </material>
>       <origin rpy="0 0 0" xyz="0 0 .08"/>
>    </visual>
>  </link>
>  <joint name="base_to_left_foot" type="fixed">
>    <parent link="base_link"/>
>    <child link="left_foot"/>
>    <origin xyz="0 0.275 0"/>
>  </joint>
> </robot>
>
> ...and I've copied the transform publisher for the transform
> /odom->/base_link that is in the Wiki. Result: My /base_link moves around as
> it should, but the foot doesn't move with it. The tf tree as shown in rviz
> shows a tree structure "/odom -> /base_link -> /left_foot", but its status
> area says that there's "No transform from [/left_foot] to frame [/odom]". I
> thought the tf tree structure did just that?
>
> Any help is much appreciated.
>
> Regards,
> Björn
>
> Am 28.12.2010 um 21:53 schrieb Chad Rockey:
>
> > I ran into a similar problem as well.  I didn't realize that
> joint_state_publisher wasn't included in ROS and needed to be installed.  Of
> course, it says right there it's needed and the warning tells you exactly
> what you need to do.  However, when you're focused on the URDF, these things
> sometimes slip by.  Maybe we should make this more obvious and explain
> robot_state_publisher's and joint_state_publisher's roles better or even put
> rosmake/rosrun tf view_frames steps in the tutorial somewhere?
> >
> > - Chad
> >
> > On Tue, Dec 28, 2010 at 1:59 PM, David Lu!! <davidvlu at gmail.com> wrote:
> > Hey Björn,
> > The tutorials were just finished last month, so they're up to date as far
> as I know.
> >
> > It looks like you have everything figured out, but I couldn't tell
> because of the last "I still need to figure out why".
> >
> > Things will appear white and at the origin when there isn't a proper
> transform to the link. If you're still running into that problem, make sure
> that rxgraph looks the way you would expect, with joint_state_publisher
>  connected to robot_state_publisher.
> >
> > -David!!
> >
> >
>
> --
> Björn Giesler
> bjoern at giesler.de
> Verschlüssel Deine Emails -- oder schreibst Du Deine Briefe außen auf den
> Umschlag?
> Meinen Schlüssel findest Du auf
> http://giesler.biz/~bjoern/downloads/bg.asc -- schick mir Deinen!
> Mehr Informationen findest Du auf http://www.gnupg.org/index.de.html
>
> _______________________________________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20101229/a4ea6ac5/attachment-0003.html>


More information about the ros-users mailing list