Abstract
The author addresses force control of robot manipulators. It is shown that from a linear system point of view the methods proposed in the literature are equivalent to state an/for output variable feedback, and the stability, for measurement performed either in joint or task space coordinates, can be established through similarity transformations. However, when selection matrices are used, the stability is affected by the choice of compensation in terms of task or joint space coordinates. Using linear models, it is argued that many different methods are analytically equivalent, hence the suggested improvements are inconclusive unless considerations of hardware-, software-, and equipment-specific data are taken into account.

This publication has 6 references indexed in Scilit: