Hi,
I am trying to simulate parachute landing human model, in which I have used Human Model from template (and Not the Standing Human) and defined postures in Mannequin.any folder. For landing, I have driven Pelvix posY vertically down. And now when I added GRF Prediction into the model, it is not working. Though it shows the foot contact nodes on both the feet. But when the foot touches the ground, the GRF values are not generated. All the foot nodes remained inactive.
Hi,
That is a very interesting simulation that you are trying to do.
In case of GRF prediction, the contact detection thresholds also check for velocity thresholds. When you drive the pelvis downwards, do you also model the part when the model hits the ground and the feet are stationary with respect to the ground?
You can check these thresholds in the settings. Further, you can also check the strength of the recruited actuators at one of the contact nodes. You can see the strength is dependent on signals detecting whether the position and velocity are within their respective thresholds for contact detection. You can try to see the evolution of these signals during the simulation to get an understanding of what is not working.
I hope that helps!
Best regards,
Dave
Thanks for the reply Divyaksh, I have tried all the possible combinations of velocity threshold. Even then it is not working. But the one which you are suggesting to make the feet stationary is not modelled. Can you help me in modeling the same.
I would like to add on to the existing simulation study that the driver acceleration (PelvisPosY acceleration which is the one I am driving), is almost zero. I used AnyEqInterpol driver and given an input to Driver Pos using PieceWise interpolation. But then I checked in the tutorial it says that the acceleration goes to zero if piecewise is used. Then I changed it to BSpline, which causes some amount of increase in GRF values but still the acceleration is negligible.
Is it possible to give Pos, Vel and acc input in AnyKinEqInterpol Driver. I have Pelvis Y segment position, velocity and acceleration from Xsens data.
One more file is added in the simulation which is ANYBODY_PATH_AMMR>/Tools/GRFPrediction/WeakResiduals.any"
This will make Hands of God as weak residuals so as to get correct GRF, I suppose.
Kindly suggest whether all the above modifications are correct?
Hi,
If you have tried playing with velocity threshold, then I think we should look somewhere else. Just to be sure, please also check the strength of the recruited actuators of the GRF prediction so you are sure that the actuators have some strength and the contact conditions are satisfied when you expect them to be satisfied. Other aspects to check can be residuals. The Human template that you started with adds reaction force at the pelvis. This means GRF prediction won't be used as the solver will prefer the "free" reaction force over the "costly" recruited force of GRF Prediction. You should add the Weak Residuals file as you mentioned (make sure the code is included in your study). But you would still need to disable the reaction force at the pelvis. You should also check if you have reactions on at your pelvis driver.
You mentioned you have data from Xsens. Is it not full body data that you could perhaps use directly into one of the mocap models? If you only have the pelvis data, then I think interpolation driver is alright. You can use the type Bspline and then it should calculate the accelerations for the segment. You can only provide the position data in the interpolation driver.
There are different ways in which you can control position of the feet. If you are driving the pelvis with an interpolation driver, you could perhaps set the last few steps to be at constant position. You could also consider adding a kinematic limit to the movement of the feet, so it doesn't penetrate into the ground. You can see, for example, how our joint limits are set up. This is a little bit complex as you would need to make sure the leg has weak drivers at the joints.
I hope that helps.
Best regards,
Dave