Why ground_plane has mu 100 mu2 50 if documentation says it's in range [0..1]

Checking out friction parameters trying to solve situations involving grasping and a humanoid walking I ended up with a doubt:

If I check mu and mu2 parameters from ground_plane in the different sdf versions they have values like 100 and 50 or 1000 and 500.

Meanwhile sdf documentation (http://sdformat.org/spec?elem=collisi...) says it's in range [0..1].

I've tried spawning an inclined box as a ramp and spawning a cube over it and play with it's mu and mu2 values. I found out that anything over 1.0 seems to be treated as just 1.0.

I've tested this on Gazebo 2.2 and Gazebo 4.0.

Here you can see a video of the friction acting like that with a tool I made to spawn a inclined ramp and a cube over it with the parameters I tell it on the fly: https://youtu.be/I_QM5v0xT14

edit retag close merge delete

Sort by ยป oldest newest most voted

I would say that's an error in the documentation, since friction coefficient can be greater than 1 depending on the surface (it shouldn't be negative though).

ODE takes the minimum of the two surface friction coefficients when objects come in contact, so I think the rationale is that the ground plane should have a large friction coefficient so that it isn't a limiting factor, since a lower friction coefficient can be accommodated by specifying a smaller value on the bottom of a robot foot, for example.

You can also set the gravity vector to have a non-zero lateral component to simulate an inclined plane. That's what we do in this friction test world.

Did that make sense?

more

Yes it did, altho in my tests any value over 1.0 created the same results than 1.0. Maybe on very big (or very small) weights it behaves different. Thank you very much!

( 2016-04-19 09:45:39 -0500 )edit

well, ODE uses the surface with the smallest friction coefficients. Did both models have large friction coefficients in your testing or just the ground plane?

( 2016-04-26 12:41:18 -0500 )edit

After learning discussion above, I want to make it clear, whether it would be effective if I set the 'mu' value of my robot wheels to 0.7 and that of the ground to a large value, then it would be meaningful for me to simulating the real situation, where the wheels and the ground have a friction coefficient of around 0.7.

( 2016-05-21 02:08:12 -0500 )edit

@scpeters, what does ground planes friction of mu1 = 100 and mu2= 50 mean? That my robot will have trouble navigating due to friction or vice versa?

( 2016-05-23 04:31:44 -0500 )edit