A new method for the calibration of multi-axis, linear and redundant force sensors is presented. This new procedure, named device hyperplane characterization method, is inspired by the shape from motion method for it reduces the burden represented by the need for a huge number of force measurements, typical using least-squares methods, in order to reject errors during the calibration procedure. The proposed technique is an application of the rank theorem and achieves good noise rejection without much time consumption focusing on sensor output measurements, and reducing the effect of disturbances operating the projection of raw output data on the hyperplane to which measurements are ideally compelled to belong in the case of redundant sensors.