Keep a joints position fixed

[geerten]

I was wondering if it is possible to keep the position of a joint fixed.

If I have a revolute joint simulating an axis, in real life this is also fixated (and only able to rotate around the joint itself). However, in V-REP when enough force is applied, the blue part of the joint will move from its original position.

Is it possible to keep the joint fixated at one position and really only allow movement along the DOF of the joint itself?

[coppelia]

Hi Geerten,

What you are describing happens when your dynamic simulation set-up is badly designed. Physics engines like Bullet or ODE work with constraints to model joints, and if applied forces are too large (relatively), the constraints can break apart. Try to adjust the masses and inertias for the bodies on one side and the other of the joint. And always follow the design considerations related to dynamics.

Cheers

[tdewolf]

I'm also running into this problem, I've followed the instructions for making dynamic joints, but I'm trying to model an actual robot and I have the weights. Is there no way to force a joint to stay aligned with its axis when there's a lot of torque placed on it? I can make things lighter but it makes the simulation less useful.

[coppelia]

Depending on the used physics engine, you often don't have the choice except to adjust or balance out the masses and inertias. But you will notice that with Vortex and Newton, those requirements can largely be loosen, and realistic masses/inertias be used.

Cheers

[tdewolf]

Interesting! Could I hear more about what requirements should be loosened, or a link to a page talking about it?
Thanks!

[coppelia]

This page gives a good overview. Mainly the design considerations 7 and 8.

Cheers

[tdewolf]

Thanks!