Asti model - how does hybrid mode affect IK

[nex]

Hello,

I am wondering how the inverse kinematic mode (with hyprid operation checked) in the Asti robot works.

When you put Asti in an empty scene, and remove the friction of the floor, the feet slide along the path defined at the feet (leftFootPath/rightFootPath) as you would expect: the feet move on the floor in a kind of infinity symbol like manner.

What I do not understand is how blocking the feet from sliding (by leaving the friction intact as it is in a new scene) backpropagates those forces to the hips. I get that the part of the paths, where the feet are still touching the ground, cannot exert force on the feet since they are held back by friction. But how does IK with hybrid operation backpropagate this force up to the hips? Shouldn't the friction just cause the force to get "lost"?

[coppelia]

Hello,

the hybrid IK mode works as follows:

• Basically Ik will be computed for each leg. We know the desired foot position/orientation, and we compute the leg joint values that satisfy this. This doesn't involve dynamics at all.
• The calculated joint values of the joints in hybrid IK mode will be applied as desired position, where the joints in hybrid IK mode are actually in force/torque position control.

So basically the leg joints are all controlled in position, and the target joint positions (i.e. desired joint positions) are provided by the Ik module.

Cheers

[nex]

Thanks, I understand now.

So the foot position itself isn't really controlled, it is just an input for the IK engine that computes the joint positions assuming the foot can be freely positioned. When the joint positions are then indeed realized by the position controller, it does not care where the foot actually will be. It just applies any forces it needs to reach the target joint positions. This will cause the hip to move when the foot is on the ground and cannot move due to friction, because the desired pose has to be realized by "backpropagating" up the body.


How did you design the path for the feet? I think it's not always easy to imagine their final effect (unconstrained vs. constrained due to, for example, friction).

[coppelia]

I think the path was designed by trial-and-error, keeping in mind to have the center of mass always above one of the two feet, for a quasi static motion.

Cheers