waffle_pi cartographer failing about turtlebot3_simulations HOT 10 CLOSED

It seems i indeed forgot to update the turtlebot3 package - now everything runs as expected.

Thank you for your efforts ;)

from turtlebot3_simulations.

Hi,
It looks like the Rviz has some issue with filtering the IMU sensor data.
Could you turn off the imu data option and see if it fixes the issue?
https://github.com/ROBOTIS-GIT/turtlebot3/blob/foxy-devel/turtlebot3_cartographer/config/turtlebot3_lds_2d.lua#L52

from turtlebot3_simulations.

Yes, turning off the imu data option clearly helps. But is the IMU sensor data different between the burger and the waffle_pi model?

from turtlebot3_simulations.

No, both Burger and Waffle Pi use the identical IMU as they are using the IMU on OpenCR1.0 board.

from turtlebot3_simulations.

But where is the difference then? Burger model is running fine with IMU data option turned on while waffle_pi is not?

from turtlebot3_simulations.

There was a missing imu joint in the sdf file and this PR(#133) will fix the issue in the gazebo.

from turtlebot3_simulations.

After seeing that you merged the PR into branch ros2-devel, i switched to this branch, pulled (i did see the change in the 4 sdf files), turned IMU data back on in the cartographer config and ran my usual setup:

1. ros2 launch turtlebot3_gazebo turtlebot3_world.launch.py
2. ros2 run turtlebot3_gazebo turtlebot3_drive
3. ros2 launch turtlebot3_cartographer cartographer.launch.py use_sim_time:=True

in parallel terminals. While the bot moved very well and avoided objects as intended before the change, it now starts after a short time (usually under one minute) to move backwards, collide with the walls and flip over and over.
If i switch to the burger model in all 3 terminals and run the same commands it still behaves as expected.

from turtlebot3_simulations.

@GrueneNeun
Thanks for the update.
I'm currently checking on the Waffle Pi. I'll update the package when the issue is resolved.

from turtlebot3_simulations.

After this commit the waffle_pi behaves nearly as in the beginning - it avoids objects as expected, but the cartographer does the same flipping as before. Turning off the IMU option in cartographer still helps.

from turtlebot3_simulations.

@GrueneNeun
Thank you for the feedback.
I'll run the simulation again and see if I can reproduce the problem.
Happy New Year!

=========Appended===========
I just ran a good 15 minutes cartographer SLAM with the latest foxy-devel commits of turtlebot3 and turtlebot3_simulations packages, but wasn't able to reproduce the issue.

Could you check if your packages are up to date, especially the turtlebot3 package?

from turtlebot3_simulations.