DART contact always sets mu2 = 1?

edit

Hello Gazebo-land,

Very quickly:

Here's what I am running:

• Gazebo 11.0.0~pre1 on Ubuntu 18.04
• Compiled from source according to http://gazebosim.org/tutorials?tut=install_from_source&cat=install
• Included DART support according to install instructions

Here's my .world file:

<?xml version="1.0" ?>
<sdf version="1.7">
  <world name="default">
    <physics type="dart">
      <max_step_size>0.001</max_step_size>
      <real_time_factor>1</real_time_factor>
      <real_time_update_rate>1000</real_time_update_rate>
      <dart>
    <solver>
      <solver_type>dantzig</solver_type>
    </solver>
    <collision_detector>dart</collision_detector>
      </dart>
    </physics>

    <!-- Ground -->
    <include>
      <uri>model://ground_plane</uri>
    </include>
    <!-- Cardboard Box -->
    <include>
      <uri>model://cardboard_box/</uri>
      <pose>0 0 1 0 0 0</pose>
    </include>

  </world>
</sdf>

Notice that I intend to use DART for physics and collision detection. The model referred to is the standard Gazebo model called cardboard_box. (I have not included it here.)

Within models/cardboard_box/model.sdf we set mu = mu2 = 0.0

  <friction>
    <ode>
      <mu>0.0</mu>
      <mu2>0.0</mu2>
    </ode>
  </friction>

The problem: Once the box is sitting on the ground, we can play with gravity in order to effectively change the inclination of the ground. Two observations:

1. [For example] If we set gravity(in the order {x,y,z}) to {1.0, 0.0, -9.81}, then the box takes off in the +x-direction due to the frictionless surface.
2. If we set gravity to {0.0, 1.0, -9.81}, the box does not move. In fact, the box will not move for any value for the y-dir gravity below 9.81.
3. If we set gravity to {0.0, 9.82, -9.81}, the box will start to slowly slide in the +y-direction at a much reduced acceleration.

This behavior suggests that mu2 is being set to a value of 1, regardless of what is in the sdf file.

I'm trying to figure out what is going on, what I'm missing or have set up incorrectly, or if there is a bug.

Thanks for the help.

-GNelson