The short answer: It doesn't know. Longer answer: The force of gravity is always a constant down vector relative to the robot. Since its constant, the counteracting force it needs to apply to remain static is independent of any other forces applied to it. It is always adjusting to gravity and only gravity. A cool thing about that is that, since it is always counteracting gravity, i assume the person moving weight around feels like the weight is floating in space. Not weightless, but weighless with respect to gravity. must feel weird.

I don't think it's measuring force at all, at least not during the video shown. As long as it knows the angle that each joint is at, and the density of the material that it's made out of, it could calculate the gravitational force felt by each segment, and counteract only that force. For more practical accuracy, it could have a "Calibration Mode", where it attempts to "Hover" the tip of the mechanism at various x/y/z points, in order to "measure" how much force it needs to apply at each x/y/z to maintain a hover at each x/y/z (And could extrapolate for x/y/z values that are in-between one of the actually calibrated x/y/z points). Then it could adjust its internal theoretical model to the actually measured results, for more fine-tuned accuracy (Just like how google maps will often ask you to calibrate your gyroscope, for improved accuracy). E.g. the calibration mechanism could help make adjustments for the gravity felt by those "wires", which are harder to model from a purely theoretical basis.

"...or it pretty violently shakes while settling." That would be a pretty neat trick for a passive mechanism.

what simulation and what graphs

"translation" DOF is defined here as the distance of the cantilevered load from the last joint. The prior art is able to compensate for 2 DOF when the load is held near the joint, but fails when the load is moved away from the joint. The new technique seems to be able to compensate for this.

@@denniszhang9278Its a bit of a stretch but yeah makes sense👍

@@FullFledged2010 It's not a stretch at all. This is how manipulator kinematics are defined. There are two types of joints: rotational (R) - a rotation around a single axis, and Translational (T) - translation (movement) along a single axis Number of linearly independent joints gives you DOF, so in this case we have a RRT kinematic arm. (3 DOF)

@@Axymerion roll,pitch,yaw,heave,surge,sway. Are the 6 degrees of freedom. This system only has roll and pitch. . And technically you have surge but in a very limited way🤔 pitch and surge are coupled so its not really degrees of "freedom" now is it?

@@FullFledged2010 You forgot about the weight, which is a translational joint (in a vertical direction). It can move independently of the other two joints, and thus is the third DOF

Weight position on the final arm