[ros-users] TrajectoryPlannerROS does not take parameter

[Eitan Marder-Eppstein]

Christian,

First off, thanks for trying to help reproduce this again. When you say that
you updated the navigation stack to version 1.0.7... what version did you
update from? I think your previous e-mail referenced 1.0.7 as the version
that you were running from the outset. This might help me track down whether
this could have been a bug in previous versions of the navigation stack. It
still seems unlikely to me, since the code responsible for simulating
trajectories hasn't changed in awhile, but its probably worth a look.

As far as the parameters go... I've tried to configure the navigation stack
using the parameters you posted when you were having trouble and couldn't
get things to fail. The low acceleration limits never caused the velocity
limits to be violated or the robot to move backwards at high speed.

Really, really strange... we may just have to let this one go in the end. I
hate weird, unexplainable things like this.

Hope all is well,

Eitan

On Wed, Jun 30, 2010 at 2:36 AM, Christian Verbeek <
verbeek at servicerobotics.eu> wrote:

> Dear Eitan,
>
> I did some more tests trying to reproduce the behaviour of driving
> backwards. I failed.
>
> The time I increased the rate of the /scan topic to 6Hz everything
> worked well. Now I lowered the rate to 3Hz to test, if this was the
> reason for the weird behaviour I saw 2 weeks ago. Even with this low
> scan rate and lots of CostMap timeouts the planner works well.
>
> I modified a couple of things in the meantime. The local planner
> parameters are now very similar to the default values. I updated the
> navigation stack to 1.0.7.
>
> I am glad its working but I would also like to know why it was not
> working right from the start. The only reason I can think of after I did
> these tests today is that
>
> 1) there was a bug in the navigation stack I used 2 weeks ago and this
> bug has been fixed
> 2) the local planner parameters I posted here lead to the poor
> performance (I had very low acceleration values)
>
> Regards
> Christian
>
> _______________________________________________
> 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/20100630/e36efa99/attachment-0003.html>