Hello, I'm using Gazebo to simulate the dynamics of a set of curved entangled tubes, to emulate a real-life experimental setup. Geometrically, I'm representing the curved tubes as a linked list of cylinders with fixed joints, since they are made of rubber and are practically rigid.

My issue is that when I use the real-life values for the (links') mass, the tubes tend to slip when placed on top of another tube. If I multiply all the tubes' mass by 5, this issue does not occur and the whole system of tubes remains "stable" throughout the simulation. However, I would prefer to use the tubes' actual masses, in order for the simulation to be as realistic as possible.

My question is: What physics parameter(s) do I need to change (and to what values) to prevent the tubes from slipping when using their real-life mass?

Here are the links to my .world files, which contain all the parameters I'm using. You can run yourself these files in Gazebo to see what's happening:

• Here is the .world file with actual mass.
• Here is a .world file identical to the first one, with the sole difference that I multiplied all the links' mass by 5.

In addition, here is a summary of the physics parameters I'm using:

• Physics engine: ODE (type = "quick" and "iters" = 100).
• Mass por unit of length: 0.2 kg/m (real-life value), 1.0 kg/m (for the second .world file). Note: I'm programmatically generating these .world files.
• Moments of inertia: I'm computing them using the standard formulae for a solid cylinder (please refer to https://en.wikipedia.org/wiki/List_of_moments_of_inertia).
• Surface/friction/ode/mu: 100000.0 (same for mu2).
• Surface/friction/ode/slip: 0 (same for slip2).
• Surface/friction/torsional/coefficient: 1.0.
• Surface/friction/torsional/usepatchradius: false.
• Surface/friction/torsional/surface_radius: 0.01.
• Surface/contact/ode/kp: 1000000.0 (1e+06).
• Surface/contact/ode/kd: 1.0

I've already tried changing the values of these parameters, but to no avail.

Thank you in advance for your time and help.