Revision history [back]

This actually is a follow up of still-runaway-simulations-in-gazebo. There I thought that the solution would be to use the dart engine, but I don't know how to set coulomb friction there, which I really need. So I am back with using ODE.

Please find here the boot.sdf and instable.mp4 that show the instability (but no crashing) in the model. It contains 5 revolute joints like a shoulder and elbow. The instability as shown in the mp4 is very reproducable.

The model arms.urdf is a little more complicated, but does not exibit the instabilities. The complete model, to be found here in boot.sdf crashes almost directly.

The question is, is there something wrong in ODE or is the current setting of ODE not good for my use case. Thanks again, Sietse

This actually is a follow up of still-runaway-simulations-in-gazebo. There I thought that the solution would be to use the dart engine, but I don't know how to set coulomb friction there, which I really need. So I am back with using ODE.

Please find here the boot.sdf and instable.mp4 that show the instability (but no crashing) in the model. It contains 5 revolute joints like a shoulder and elbow. The instability as shown in the mp4 is very reproducable.

The included model arms.urdf arms.urdf is a little more complicated, but does not exibit the instabilities. The complete model, to be found here in boot.sdf crashes almost directly.

The question is, is there something wrong in ODE or is the current setting of ODE not good for my use case. case.

UPDATE: Even setting the max step size to 0.0001 does not help much.

Thanks again, Sietse

This actually is a follow up of still-runaway-simulations-in-gazebo. There I thought that the solution would be to use the dart engine, but I don't know how to set coulomb friction there, which I really need. So I am back with using ODE.

Please find here the boot.sdf and instable.mp4 that show the instability (but no crashing) in the model. It contains 5 revolute joints like a shoulder and elbow. The instability as shown in the mp4 is very reproducable.

The included model arms.urdf is a little more complicated, but does not exibit the instabilities. The complete model, to be found here in boot.sdf crashes almost directly.

The question is, is there something wrong in ODE or is the current setting of ODE not good for my use case.

UPDATE: Even setting the max step size to 0.0001 does not help much.

2nd UPDATE Getting a bit desparate, after 6 month trying to find a solution. Changeing to gazebo 10.1, from source, does not help either.

I would be helped very much to have at least some confirmation that someone has the same problem (instable.mp4) or not with my boot.sdf mentioned above.

Thanks again, Sietse