Abstract Parallel manipulators, by their architecture, have the potential for high rigidity and hence high positioning repeatability. In addition, the availability of an analytical inverse model capable of accommodating all kinematic parameters, means that very high positioning accuracy is also possible if the actual parameter values are known. The inverse kinematic model, which is in analytical form, is used to generate an analytical identification jacobian used to perform a sensitivity analysis and for the identification scheme. The calibration presented here is able to globally estimate all the parameters of the platform including the offset of the legs parameters. The identification uses a singular value decomposition of the identification jacobian. This provides an excellent immunity to numerical redundancies as may be caused by intrinsically redundant parameters or by the measurement of a subset of the full pose coordinate variables as calibration data. Extensive simulation results for 6 DOF platform are presented.
Oliviers et al. (Sun,) studied this question.