Sven,

I would expect move_base to have trouble operating on a map that has
half-centimeter resolution like the one you've passed it. Its true that the
most expensive inflation operation happens only on startup (global updates
and inflation), but there's still a lot of local updates that occur while
running: raytracing, local inflation, etc. I think that the highest
resolution map we've run navigation on is 2.5cm. The easiest solution, would
be to reduce the resolution of the map before passing it to move_base. Is
there a reason that you need to operate in such a high-res enviornment for
navigation?

Hope this helps,

Eitan

On Tue, Jul 27, 2010 at 2:26 PM, Sven Olufs <olufs@acin.tuwien.ac.at> wrote:

>  Hi there,
>
> I have a strange problem with the "move_base" module:
>
> The CPU load is always 100 % (on a Quad Core Machine !) Is this normal ?
> I can lower the control_loop_rate, but there is no difference at all in the
> runtime.
> When I lower the update rate of the local_costmap to 2Hz i get 60% load...
>
> I used the settings that are proposed in the "navigation tutorial", so
> nothing special..
>
> I also get strange Warning like:
> "Map update loop missed its desired rate of 0.5000Hz" for the
> global_costmap (6000x6000, 0.005m per pixel)
> and the global_costmap never publish anything (using rviz)
>
> The robot moves also very strange, i.e. stop-and-go behaviour, and seem to
> follow the path with a delay of 2sec !!
>
> I assume it is the global map that consumes the computational power, but as
> far i understand it is only updated once (at the start)
>
> Thanks for help in advance..
>
> Sven
>
> --
> Sven Olufs, M.Sc
> Research Fellow
>
> ACIN | AUTOMATION&  CONTROL INSTITUTE
> INSTITUT FÜR AUTOMATISIERUNGS-&  REGELUNGSTECHNIK
> TECHNISCHE UNIVERSITÄT WIEN
> Gußhausstraße 27-29 | 376. 1040 Wien
> Tel. +43 (0)1 58801 - 37666
> Fax. +43 (0)1 58801 - 37698
> olufs@acin.tuwien.ac.at | www.acin.tuwien.ac.at
>
> _______________________________________________
> ros-users mailing list
> ros-users@code.ros.org
> https://code.ros.org/mailman/listinfo/ros-users
>