simtk.org

Dear all

Does anyone know what algortim is implemented into OpenSim 3.3 for Inverse Dynamics ?

Actually,OpenSim Documentation on confluence website [1] suggests that the algoritm is based on a global optimization, described in the Kuo's paper [2]
But theoretical manual of SimBody(I have the 2013 version) cites the Jain's paper[3]

In particular I would be interested to know if it is a recurcise Newton-Euler algorithm or a step by step global approach

Thank you for help

[1] https://simtk-confluence.stanford.edu/d ... seDynamics
[2] Kuo, A.D. A least squares estimation approach to improving the precision of inverse dynamics computations, Journal of Biomechanical Engineering, vol. 120, pp. 148-159, 1998.
[3] Jain, A.; Vaidehi, N.; Rodriguez, G. A fast recursive algorithm for molecular dynamics simulation. J. Comput. Phys. 106(2):258-268 (1993).

Sorry for the confusion. OpenSim, before its full integration of Simbody (before 2.0), used to solve for the generalized forces that generated the user-supplied acceleration, iteratively.

Since then, all multibody dynamics are performed with Simbody. The InverseDynamicsSolver provides a simpler interface to Simbody for solving Tau = M*u_dot-G(q)-C(q,u)-A(q,u,t,x), where the matrices are not formed explicitly but recursively as in the algorithm by Jain.

The paper of Kuo is not a substitute for a fast InverseDynamicsSolver, it is about the experimental treatment of inverse dynamics where kinematics and externally applied forces are not independent. The external reaction forces accelerate the center-of-mass of the system and one can use the redundancy (between measured forces and estimated acceleration) to improve the accuracy (dynamic consistency) of the solution when solving joint moments. It is a technique for addressing the uncertainty in the measurements and the model. This is what RRA attempts to do in OpenSim, and the poster by John et. al. at the end of the Inverse Dynamics section of Tutorial 3 is describing that.

Thank you for making it more clear.

So that's why OpenSim is able to compute a solution even if there is a lack of some external forces (at the opposite side of the dynamic chain for example)