Robotics StackExchange | Archived questions

hector slam raspberry problem

Hello everybody,

I use to create maps Hector Slam. So far everything has worked. I will now use a Raspberry Pi 2 with ubuntu to create maps but I always get error.

Hector Slam starts without problems and Rviz connects without problems over the network with the Raspberry.

In the area of laser scanners in Rviz, is an error message which reads:

For frame [laser]: No transform to fixed frame [map]. TF error: [lookup would require extrapolation into the future. Requested time 2529.375686649 but the latest data is at time 2529.225683316, When looking up transformable from frame [laser] to frame [map]]

I think there is a performance problem, because it worked on my computer without any problem.

Here's my TF Tree

https://www.dropbox.com/s/24xdfgkmjs3he0w/frames.pdf?dl=0

Here's a screenshot of the error

https://www.dropbox.com/s/qdj71u9arvmrg84/Screen%20Shot%202015-08-13%20at%2011.39.24.png?dl=0

Has anyone any idea how I can solve this problem?

The Laser Scanner is Hokuyo UTM-30LX-EW

Thanks + greetings

Max

Asked by Twista3579 on 2015-08-13 05:04:06 UTC

Comments

Answers

This sounds like you might be experiencing a time sync problem between computers. The rate at which hector_mapping publishes tf data suggests that it either cannot keep up with the incoming LIDAR data, or there is a tf/time sync problem that causes huge waiting delays. I tested things on a RPi2 as described in MappingUsingLoggedData and CPU consumption was about 10%, so I don't think the machine is too slow normally.

Did you compile hector_mapping yourself? If so, you should specify the build type:

catkin_make -DCMAKE_BUILD_TYPE=Release

Otherwise, things might get built without optimization and can be very slow.

Asked by Stefan Kohlbrecher on 2015-08-13 07:01:40 UTC

Comments

thank you. It works. That was the solution to the problem. The CPU consumption is about 25 % instead of over 100% :-D

Asked by Twista3579 on 2015-08-15 01:32:17 UTC