Ecole des Mines de Nantes, 4 rue Alfred-Kastler, Nantes 44307, France
Institut de Recherches en Communications et en Cybernetique de Nantes, 1 rue de la Noe, 44321 Nantes, France
(Submmited to "IEEE Transactions on Robotics", 2011)
Previous page | Outline | Next page
Let us analyze in details expression that allows computing the stiffness matrix of a serial chain with passive joints from corresponding matrix of the chain without passive joints . Assuming that , this matrix can be factorised into the product of two non-singular square matrices . This yields a compact presentation of the desired matrix in the form
(A1) |
where
(A2) |
and the inverse exists due to the assumption . Further, the product can be also factorised using the SVD-technique [50] as
(A3) |
where , are the orthogonal matrices and the matrix is composed of non-zero singular values (provided that ). The latter gives the following presentation of
(A4) |
where the product may be computed in a straightforward way:
(A5) |
So, the inner part of may be presented as
(A6) |
This leads to the final expression
(A7) |
that allows to evaluate the rank of the stiffness matrix
(A8) |
and to justify Remarks presented in Section III.
For computational convenience, the orthogonal matrix may be split into six vector columns and the matrix product is expressed as a subsume of corresponding to the non-zero elements of . This gives another presentation of the desired Cartesian stiffness matrix
(A9) |
where the middle term includes only those unit vectors that are not “compensated” by the passive joints (for this notation, the directions correspond to the end-effector motion due to the passive joints, which do not produce elastostatic reactions). It should be noted that in the case of , i.e. without passive joints, the total sum of produces a unit matrix and the latter expression is reduced to .