[ros-users] Costmap2DROS timeouts and poor amcl performance

[Eric Perko]

On Sat, Dec 18, 2010 at 4:41 PM, zdydek <zac at mit.edu> wrote:

>
> Hi all,
>
> I'm having some trouble running the navigation stack on my robot. I get
> periodic Costmap2DROS transform timeouts which cause the planners to stop
> and replan. This happens about once every second leading to really poor
> performance.
>
> I ran tf_monitor and found that everything looked ok (odom being published
> at 50 Hz, little delays in all tf's) with the exception of amcl, which had
> a
> max delay of around 0.25 sec. CPU utilization is high (50-75%), but not
> locked at 100%.
>
> I tried various things to improve the performance including reducing the
> size and resolution of my costmap, and reducing the number of particles in
> amcl (all the way to 100 particles!). This seems to have helped the problem
> a bit, but the the behavior still occurs once every couple seconds.
>
> Not sure how I should proceed from here. I would really appreciate your
> help!
>
> Thanks,
>
> -Zac
>

Zac,

Could you post your current configs for move_base and amcl? It's kinda hard
to say what might be causing your problems without knowing what settings you
are trying to use.

What type of computer are you running this on? Size of a priori map? And can
you include the output of 'rosrun tf tf_monitor' (I'm interested to see the
delay information for amcl since what reported seems unusual - amcl delay
should be negative since it future stamps all of it's transforms)?

- Eric


>
>
> --
> View this message in context:
> http://ros-users.122217.n3.nabble.com/Costmap2DROS-timeouts-and-poor-amcl-performance-tp2112835p2112835.html
> Sent from the ROS-Users mailing list archive at Nabble.com.
> _______________________________________________
> 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/20101218/ef790f6a/attachment-0003.html>