[ros-users] Move_base fails
eitan at willowgarage.com
Thu Feb 3 18:38:05 UTC 2011
On Wed, Feb 2, 2011 at 10:12 PM, abhy <abhy.12354 at gmail.com> wrote:
> Hello Eitan,
> I did check topics you have mentioned. Lately i found that when i launch
> stack at first time /initial_pose is not getting published in spite of
> sending x,y,a through parameters in amcl.
Can you post the results of what you saw from the checks I suggested? The
output might actually help us figure out what's going on.
> I always have to click in rviz to set its pose and target.
You can write code that publishes on the "initialpose" topic, but that will
only work if the robot starts in about the same place in your map every time
it comes up. Typically, we just set the initial pose in rviz once when we
bring our robots up.
> in short, i have to set initial pose and then set goal at least for the
> first time from rviz and then only from the next time stack takes goal from
> the code.
Are you saying that once you set the pose of the robot in rviz things start
working for you? If that's the case then, yes, the robot must be localized
before navigation can be expected to work properly. This means you need to
set the initialpose of the robot, either from code or in rviz, properly
before you send goals to the navigation stack.
> Is there any way by which i can publish amcl pose properly at first
> to get pointcloud? so that i do not need to set goal from rviz for the
As I mentioned above, you can do it from code by sending a
PoseWithCovarianceStamped message (
to amcl on the "initialpose" topic. However, be careful with this because it
means you'll have to make sure that the robot always starts in the correct
position in the world. I'd suggest just setting the "initialpose" in rviz on
startup... or writing some code to solve the kidnapped robot problem since
we haven't gotten around to it yet... patches are always welcome ;)
Hope this helps,
> Any guidance on this would be appreciated.
> View this message in context:
> Sent from the ROS-Users mailing list archive at Nabble.com.
> ros-users mailing list
> ros-users at code.ros.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the ros-users