[ros-users] navigation stack : changing inflation_radius

[Eitan Marder-Eppstein]

Kei,

The robot has two recovery behaviors by default.

The clear_costmap_recovery: http://www.ros.org/wiki/clear_costmap_recovery
The rotate_recovery: http://www.ros.org/wiki/rotate_recovery

Note that the rotate recovery will only run when it is safe for the robot to
rotate in place. So, in the case that the robot is surrounded by people, it
may not run.  Also, there are two different versions of the
clear_costmap_recovery that run. One which clears out old obstacles outside
of a conservative radius, and one that clears out old obstacles outside of a
much more aggressive radius. For more information see:
http://www.ros.org/wiki/move_base#Parameters

If you receive the message you posted above, it most likely means that the
rotate_recovery_behavior did not think the robot could rotate in place
safely, but they should have run. Turning on the DEBUG log level on
move_base would verify this.

Hope this helps,

Eitan

On Thu, Sep 2, 2010 at 10:18 AM, Kei Okada <k-okada at jsk.t.u-tokyo.ac.jp>wrote:

> Hi,
>
> Thanks, your answers are good news for us. I'm afraid that I've chosen
> a stupid solution.
>
> As for the second question,  I send a goal pose, after the robot stops
> navigation ( which means waitForResults returns false) or the robot is
> surrounded by people, expecting that the robot will try recovery
> behaviors such as rotating. But the robot aborts the action after a
> few seconds with "Aborting because a valid plan could not be found.
> Even after executing all recovery behaviors" message, although it
> seems no recovery behaviors ware executed.
>
> So, I would like to know what happens inside.
>
> do {
>
> }
>
> On Thu, Sep 2, 2010 at 3:17 PM, Eitan Marder-Eppstein
> <eitan at willowgarage.com> wrote:
> > Kei,
> > I've answered in-line below.
> >
> > On Wed, Sep 1, 2010 at 7:55 PM, Kei Okada <k-okada at jsk.t.u-tokyo.ac.jp>
> > wrote:
> >>
> >> Hi
> >>
> >> I have two questions regarding the navigation stack.
> >>
> >> 1)  is it possible to change inflation radius after we start move_base?
> >>    Currently we set new
> >> /move_base_node/local_costmap/inflation_radius parameter and rosnode
> >> kill /move_base_node.. but this takes about 10 seconds for respawn.We
> >> look for faster solution.
> >
> > Currently, there is no way to dynamically reconfigure the costmap's
> > parameters, including the inflation radius. Handling dynamic footprints
> is
> > something that's on the todo list for navigation, but, unfortunately, its
> > not implemented yet.
> >
> >>
> >> 2) The robot behaves according to the move_base Default Recovery
> >> Behaviors shown in http://www.ros.org/wiki/move_base page
> >>   Is there a way to know current robot behavior (navigation?clarning
> >> rotatoin?)
> >
> > If you turn the DEBUG logging level on for move_base, you'll at least see
> a
> > print out of "In clearing/recovery state" when a recovery behavior is
> > executed and "In planning state" when planning is being executed, but
> > there's no great way to tell programmatically. I've never had a request
> to
> > make that part of move_base's state external. I could probably add it to
> the
> > feedback given from the MoveBase action in a future version of
> navigation.
> > If that's something you'd like to see, feel free to ticket me on it. If
> you
> > really wanted to, you could write your own recovery behavior plugins that
> > notify you when they're active and use those instead of the defaults.
> > Hope all is well, and sorry that both the answers were probably not what
> you
> > were hoping to hear,
> > Eitan
> >
> >>
> >> _______________________________________________
> >> ros-users mailing list
> >> ros-users at code.ros.org
> >> https://code.ros.org/mailman/listinfo/ros-users
> >
> >
> > _______________________________________________
> > ros-users mailing list
> > ros-users at code.ros.org
> > https://code.ros.org/mailman/listinfo/ros-users
> >
> >
> _______________________________________________
> 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/20100902/8b1d6bb6/attachment-0003.html>