[ros-users] Standardization of mapping interfaces

Mani Monajjemi mani.monajjemi at gmail.com
Mon Nov 5 22:17:58 UTC 2012


Hi,

I've noticed that, many
<https://www.google.com/search?q=The%20doc%20job%20status%20for%20is%20currently%20UNSTABLE>wiki
page<https://www.google.com/search?q=The%20doc%20job%20status%20for%20is%20currently%20UNSTABLE>
s<https://www.google.com/search?q=The%20doc%20job%20status%20for%20is%20currently%20UNSTABLE>generated
by the new system has
this error in their headers:

*The doc job status for <package-distro> is currently UNSTABLE. See jenkins
for details.

*I went through the jenkin's console outputs and I found different errors
for different packages. Is it the task of package maintainers to solve the
problems or is there any problems with the new system? Some errors are
weird, like "


[rospack] Error: package/stack <package> depends on non-existent package roscpp

"
*
*Best,
Mani

Mani Monajjemi


On Thu, Oct 4, 2012 at 10:49 AM, Stéphane Magnenat <
stephane.magnenat at mavt.ethz.ch> wrote:

> Hello,
>
> I am working on the release of our modular ICP pipeline [1] (2D and 3D),
> which provides a mapping-style interface. While doing so, I realized that
> some features are commonly available in mapping packages, like getting the
> latest version of the map or reseting it. However, I have not find any
> standard to do so. Worst, while for 2D there is a message nav_msgs::GetMap,
> there is no standard equivalent in 3D. To allow easy interoperability
> between packages doing the same thing, it is important to have these.
>
> So here are my suggestions:
>
> - To add a GetPointMap message similar to GetMap but taking a PointCloud2
> instead of an OccupancyGrid. This could either be added to nav_msgs or to
> something new like nav3d_msgs or map3d_msgs within common_msgs.
>
> - To define a standard interface for interacting with mapping packages. If
> we consider existing packages, (for instance gmapping, hector_mapping),
> there is a "dynamic_map" service that returns the latest map. While this
> naming is strange, it is the closest thing to a standard that we have.
> Should we extend it in 3D to have a dynamic_point_map service? Also, should
> we generalize the "syscommand" topic, which is a bit awkward, or maybe add
> something like a "reset" service? We might also want to save and load maps.
>
> Comments welcome, I think it would be great if we could improve the
> interoperability of mapping packages.
>
> Stéphane
>
> [1] http://www.ros.org/wiki/**modular_cloud_matcher<http://www.ros.org/wiki/modular_cloud_matcher>
>
> --
> Dr Stéphane Magnenat
> http://stephane.magnenat.net
> ______________________________**_________________
> ros-users mailing list
> ros-users at code.ros.org
> https://code.ros.org/mailman/**listinfo/ros-users<https://code.ros.org/mailman/listinfo/ros-users>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ros.org/pipermail/ros-users/attachments/20121105/407bd384/attachment-0004.html>


More information about the ros-users mailing list