This is Srikanth from Providence, USA. I would like to know the answers for following questions.
Does Kinematic Analysis compute the position, velocity and acceleration of joints as well as deformation of muscles?
How are the forces calculated from position? Using Force Vs Length and Force Vs Velocity curves? How are these curves obtained? What if loading changes in the experiment which is used to get these curves? Could you please explain the inverse dynamics approach?
Is an inverse dynamic problem applicable only for one load case(one human type for a given load; for instance, good for running but not for walking with a backpack)
Yes. After you have performed a kinematic analysis, open a ChartFX view and browse down to the joint in question in the tree. When you open the joint, you find the properties Pos, Vel and Acc. They are the position, velocity and acceleration of the joint angle respectively.
It is hard to explain inverse dynamics in full in just a few sentences. Basically a set of equilibrium conditions determine which muscle and joint forces are necessary to obtain equilibrium. The equations are under-determinate, i.e. they have infinitely many solutions, so the right solution is chosen by an optimization algorithm. This paper explains it in more detail:
Rasmussen, J., Damsgaard, M. & Voigt, M. (2001): Muscle recruitment by the min/max criterion - a comparative numerical study. Journal of Biomechanics, vol. 34, no. 3, pp. 409-415.
- Inverse dynamics is useful for a wide range of situations and load cases. The limitations are:
a. The motion must be skilled and voluntary; crashes, slips, falls and so on cannot be analyzed.
b. The motion must be not too fast, but there is no clear limit how fast is too fast. Any sort of gait and probably also running is probably ok. A baseball pitch might be too fast, but this has not been properly investigated.
There is no limitation on the case. Backpack carrying can be handled just as well as gait or lifting or sitting or anything else you can think of.
Thanks a lot for your prompt response.
So the Kinematic Analysis sounds more like Inverse Kinematics, correct? Based on the motion of the end effector (say from motion capture), it is calculating joint angles.
So it sounds like AnyBody is solving the dynamic equations of motion, in an optimization setting. Could you please let me know if AnyBody uses Lagrangian technique to form the equations of motion?
I am wondering if, what AnyBody does, could be done by a FEA software, by modeling muscles with appropriate elements. If not, could you please point any difference in the FEA approach?
Also, I am looking to model a control problem, so that the output from inverse dynamics solution could be fed to a forward dynamics routine to compute the current position, and use PD control to minimize the error in motion. I believe AnyBody cannot handle this, please let me know if I am missing anything here.
Correct. AnyBody actually does not distinguish between forward an inverse kinematics. The system handles either of them and any combination of the two. It also copes with any level of closed chain kinematics.
AnyBody uses the principle of virtual work to form the equilibrium equations. The details can be found here: Michael Damsgaard, John Rasmussen, Søren Tørholm Christensen, Egidijus Surma, and Mark de Zee (2006): Analysis of musculoskeletal systems in the AnyBody Modeling System. Simulation Modelling Practice and Theory. Volume 14, Issue 8 , November 2006, Pages 1100-1111. Elsevier, ISSN: 1569-190X.
You can in some sense compare AnyBody’s muscle recruitment to a finite element solution. If the muscles were recruited according to a minimization of the elastic energy in the muscles and the muscles could pull as well as push, then you would come to the same solution. However, AnyBody uses different criteria and muscles are restricted to tensile forces, so you cannot obtain the same with a finite element model.
AnyBody currently does not have forward dynamics. However, some types of problems where motion is unknown can be solved with the built-in optimizer. Please refer to the tutorials on parameters studies and optimization for details.
could i use emg data to drived the Muskeloskeletal model in anybody to caculate the joint angle and joint forces?
you saying ‘‘AnyBody currently does not have forward dynamics’’, so i want to confirm it again
The answer is “no”, still not possible.
many thanks for your answer!