Force Sensor Rigidity.

[wsh1979]

I am currently building a robot with a gripper on the end. I have read about the challenges involved with gripping, and at this point I am only interested in putting the gripper on the robot. I noticed however when I parent the gripper to the robot the link is not rigid. I am using the method of putting a force senor in between to link the two together. When the joint is static and held in any orientation it appears to be rigid but when in motion it doesn't. I have attached a scene file to illustrate this. When I parent the wrist plate of the gripper to the non-respond-able static mesh (that is only there for the visual) I get the closest to it being rigid. When I parent it to the non-static respond-able convex hull the joint is far from rigid. I thought it had something to do with the weight of gripper so I built a test setup in VREP that was done pretty much the same way except with heavy (heavier than the gripper) cuboids. The result is a rigid link as I would expect. I'm not sure why the one joint in the gripper is not rigid but, I surmise something about my gripper is fundamentally making the joint behave strangely.



Attached is a scene file demonstrating this phenomena. There is a static version showing how it can all look stable if it isn't moving. There are three other setups showing the two examples of parenting and how it compares to a similar setup with cuboids. In order to see how the joints perform each setup is placed slightly off the ground to induce movement.

https://drive.google.com/file/d/0B-n2gt ... sp=sharing

Thank you very much

[coppelia]

Hello,

yes, this is a normal behaviour when running the Bullet or ODE engine. See the design considerations 7 on this page for details. Try to switch to the Vortex engine: you will see that the Vortex engine handles the situation much better than the other engines.

What you could do is balance the masses somewhat, and check also the design considerations 8.

Cheers

[wsh1979]

Hello,

What you said makes a lot of sense, however, it doesn't explain how when I intentionally increase weights using cuboids, (as in my example file) I get rigid joint behavior like I would expect. It appears that I only get this soft joint phenomena when I use the automatically generated convex hulls.

Regards

[coppelia]

Yes, beware: when you generate convex hulls or convex decomposed shapes, the mass and inertia properties of those objects does not represent anything related to that shape!
You will have to adjust it yourself.

Cheers