Unstable simulation of joint limits for a hinge joint

[simon lecleach]

I've been using warp for a few weeks and it's definitely a very impressive, well executed project!

Currently, I am having trouble simulating a hinge joint using warp.sim.
I set the joint range to "-200 45" (in degrees), the rotating link is correctly stopped by the 45deg limit.
But when I rotate the link towards the -200deg limit the simulation breaks as soon as it gets close to -180deg.

Also I have created a minimal working example here:
slecleach#1
You can easily replicate the plot above by running this file:
https://github.com/slecleach/warp/blob/minimal_example/examples/env/env_rotor.py

I am not sure what is causing this, I would appreciate if you could share some insights about this phenomenon.
Thank you!

[Eric Heiden]

Hi Simon,

Thanks for reporting the issue! There seems to be a problem in handling joint coordinates outside [-pi, pi] in general for integrators relying on maximal coordinates (XPBDIntegrator (used by default), SemiImplicitIntegrator). The FeatherstoneIntegrator handles these joint limits correctly, could you work with this for now by setting

integrator_type = IntegratorType.FEATHERSTONE

in the Environment static variables? We will investigate this issue further in the meantime.

[simon lecleach]

Hi Eric,

Thanks a lot for getting back to me! Yes we'll try the new Featherstone integrator, with minimal coordinates, this type of constraint is somewhat easier to handle.
I'll also take a look at the XPBD code to see if I can spot the issue. I'll let you know if I find anything ;)