[ros-users] Inconsistency between move_base and exploration

Michael Ferguson mfergs7 at gmail.com
Tue Jul 31 03:18:51 UTC 2012


I'm not actually aware of many (any?) people who use exploration. Most
people just teleop the robot around once when they move it to a new
environment and build the map off that -- so it is very possible that a bug
exists in such software and hasn't been noticed. Also, this doesn't really
"come with the pr2" -- it's developed by an external group (bosch R&D) and
was developed some time ago (and so something might have evolved elsewhere
in the system that causes issues).

-Fergs

P.S. -- this type of question really is most appropriate for answers.ros.org,
where you originally posted it. If someone hasn't answered there, it is
likely that nobody has the answer rather than people are missing the
question there.


On Mon, Jul 30, 2012 at 5:24 AM, Cengiz Kaygusuz <cemkyg at gmail.com> wrote:

> I've been trying to get turtlebot to do exploration on gazebo environment.
> Somehow I managed to make this work but it looks like there are still quite
> bit to be done.
> My problem is, graphically this: http://imgur.com/a/lESsv (odd pictures
> are from explore/map, evens are from /map which is constructed by gmapping)
> It looks like explore node is publishing goals according to its own map,
> while move_base are trying to execute these goals according to the map
> given by gmapping. Everything works just fine while these two are in
> agreement, but when explore publishes a coordinate which possibly conflicts
> with the map (like publishing a goal directly into an obstacle, to a
> coordinate which robot's footprint intersects with a nearby obstacle or to
> a unexplored area).
> I thought maybe it was behaving like this because of misconfiguration, but
> the same thing happens with pr2's exploration stack :
> http://imgur.com/a/y9hkL
>
> Now, my reasoning is; the exploration stack looks like the de-facto stack
> for executing exploration duties on ROS so I'd expect it to contain some
> bugs here and there but this kind of major design flaw passing by unnoticed
> is not possible.
> Another possible source of this problem may be misconfiguration but I get
> the same result with pr2 and its exploration stack yet it is again very
> unlikely that anything comes with pr2 has an unnoticed fault of this
> caliber, yet the examples given under explore_stage package is working just
> fine. I need to make things perfectly working together but from now on I
> have no idea how to proceed other than delving into some code.
>
> Thank you for your attention.
>
> - Cengiz
>
> _______________________________________________
> 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/20120730/1adc44e0/attachment-0004.html>


More information about the ros-users mailing list