[Simulation] Gazebo ROS control crashes the robot

[stevendes]

Summary

Using Gazebo ROS Control with the current robot configuration sometimes crashes the robot under specific movements, high values for the PID, or high-velocity commands being sent.

Different mass values improve the behavior, indicating that it could be an issue with the Inertia values and how Gazebo ROS control interacts.

Definition of done

• Gazebo ROS control works as expected.

[francocipollone]

One thing to try for isolating the issue is tryting the diff drive plugin without ros control to see if is an issue introduced by gazebo_ros_control or if it is originated by the robot description.

[stevendes]

You mean the Gazebo Diff Drive plugin, right? I already test it with it, and it works fine, and the robot physics also works fine when you are not loading gazebo_ros_control.

[francocipollone]

You mean the Gazebo Diff Drive plugin, right? I already test it with it, and it works fine, and the robot physics also works fine when you are not loading gazebo_ros_control.

Ah interesting. Then I would say that if this issue isn't solvable and we can confirm that it is an issue within ros gazebo control maybe we could offer the simulation to have two different setups:
1 - Using gazebo diff drive plugin by default
2 - Using gazebo_ros_control

To be selected by using a launch argument when starting the simulation.

In a way that if necessary the regular gazebo diff drive plugin can be used and we have a fully operative simulation.

[stevendes]

I'm not sure if the issue is solvable, but adding gazebo_diff_drive as an option is not hard, I'll tag you in the PR when I have it ready.

[francocipollone]

Closing as noetic isn't supported anymore