Parallelogram Linkage Issues

[William-Brumble]

I’m trying to set up a parallelogram (four bar) linkage using 6 hinges and a slider/distance constraint to form a rigid robot arm.

With the default settings (10 velocity, 2 position iterations), I can’t get rigid like behavior the hinges accumulate positional error and the linkage ends up looking loose or “floating.”

If I crank up the solver iterations, I can get the stiffness I’m after, but the CPU cost introduces noticeable frame rate jitter.

Adding damping to the joints helps a bit, but it makes the system feel springy rather than solid, which isn’t ideal for a robot like mechanism.

Has anyone run into this before or found a good way to handle closed loop linkages like this with Jolt Physics?

Thanks.

[jrouwe]

It's not entirely clear to me how many bodies you're actually using. Is it this? (assuming you're using the slider instead of a distance constraint)

I have never tried a configuration like this, but the first thing I would check is if you have properly disabled collision between the rigid bodies that intersect (Jolt doesn't do that automatically). This is one of the main causes of instability.

The next thing to check is that all of your bodies roughly have the same mass. E.g. it's easy to make 2, 3, 5 and 6 light, but that makes it very hard for the solver.

[William-Brumble]

Wow... crazy fast response broskie!

Apologies, for not specifying the bodies, but what you pictured is pretty spot on.
I've turned off the collision between the bodies but have indeed set the masses of the 2, 3, 5, and 6 to be pretty light.
I'll make them all the same mass in the morning and report back if that was a winner.

Also, tysm for this library, pretty darn epic!

[William-Brumble]

@jrouwe

That was it!

I was giving 5 and 6, relatively, a very small amount of mass. Once I bumped it up, it's acting well with the default number of position and velocity steps, thanks again!