AMCL error: Couldn't determine robot's pose associated with laser scan

asked 2014-09-20 03:02:34 -0500

IvanV gravatar image

updated 2014-09-21 06:09:28 -0500

Hi all,

I'm having a problem with AMCL in groovy, giving constantily the error "Couldn't determine robot's pose associated with laser scan".

It also is giving repeatedly a warning:

[ WARN] [1411296723.445692688]: Failed to compute odom pose, skipping scan (Unable to lookup transform, cache is empty, when looking up transform from frame [/base_footprint] to frame [/odom])

and I get this warning when using topic /initpose:

[ WARN] [1411296739.805780034]: Failed to transform initial pose in time (Lookup would require extrapolation into the future.  Requested time 1411296739.805698007 but the latest data is at time 1411296739.737616394, when looking up transform from frame [/map] to frame [/base_footprint])
[ INFO] [1411296739.805813654]: Setting pose (1411296739.805799): -0.032 0.007 0.052

AMCL is feed with the laser scans from two different sources, that are merged in a single topic using the relay from topic_tools. When fusing topics, their frame_id is not changed. Laser scan topics come from a first computer, while navigation stack runs in a diferent computer, but both are time synchronised with chrony.

The point is, even if I'm constantly getting the error message, the localization seems to be working pretty good.

The TF tree seems good (I would attach the output from rosrun tf view_frames but I can't yet attach files) and laser scans are visualized correctly in rviz in all fixed frames (map, odom, base_link...). However, after a second watch I see that odom tf is published at only 10Hz, while the lasers are published at a rate of 25Hz (possible cause?)

Altough localization seems to work, getting the error means that there is something around there that is not working properly, but I'm unable to identify the source of the error. I'm afraid also that it can be the source of other apparently unrelated problems (i.e. the second warning shown, that causes that when setting /initpose, sometimes AMCL is not getting the sent position correctly).

Any ideas of what can be causing this AMCL error?

Thank you and best regards,

Ivan.

Edit 1: Added some warnings that I'm also getting. Plus something I noticed about the publication frecuency of the different TFs.

edit retag flag offensive close merge delete

Comments

There should be some more errors that go along with this; perhaps something like "Failed to compute odom pose, skipping scan (%s)". Can you edit your question to include those as well, or a more complete log from AMCL?

ahendrix gravatar image ahendrix  ( 2014-09-21 04:23:19 -0500 )edit

I haven't find any other error, but I've checked the logs and I'm getting also a warning (I've updated the question). Also, I have noticed that lasers TFs are published at much higher frecuency than odom TF. Maybe some of them are being discarded because they have no timestamp matching odom TF?

IvanV gravatar image IvanV  ( 2014-09-21 06:11:21 -0500 )edit

Did you fix this error, if so can share how?

Kishore Kumar gravatar image Kishore Kumar  ( 2016-01-31 00:14:24 -0500 )edit

Sorry, didn't find the source of the error in the end. Everything seemed to be working fine and I didn't have time to spare looking for this. The project ended more than one year ago and I haven't looked back into it.

IvanV gravatar image IvanV  ( 2016-02-01 01:39:28 -0500 )edit