Have you tried rotating the robot (even if by nudging it yourself) a few degrees to either side of the heading it stopped at. Does that clear out the marked pixels? I've seen some pixels remained marked with high-resolution costmaps and a laser such as the Sick LMS291 that do pings that have 1 degree of angle between them.

- Eric

On Wed, Nov 17, 2010 at 5:43 PM, zdydek <zac@mit.edu> wrote:

Hi all,

I'm running into an issue with costmap_2d, not sure if it is a bug or a
parameter that is not set correctly...

Essentially some obstacles do not clear out of my costmap after they have
been moved. More precisely, when an obstacle moves and there is no farther
away obstacle behind it (within the laser range), it is not cleared form the
costmap.

For example, if the robot is driving straight down a long hallway and I
stand directly in front of it, an obstacle appears on the costmap. However,
that obstacle is not removed after I leave, making it difficult or
impossible to continue navigating down the hallway.

My laser scanner is set with "marking: true" and "clearing: true" in my
costmap_common_params.yaml.

Any idea where the problem may lie?

Thanks very much,

-Zac
--
View this message in context: http://ros-users.122217.n3.nabble.com/Clearing-old-obstacles-from-costmap-tp1920291p1920291.html
Sent from the ROS-Users mailing list archive at Nabble.com.
_______________________________________________
ros-users mailing list
ros-users@code.ros.org
https://code.ros.org/mailman/listinfo/ros-users