Periodical in-situ re-calibration of force platforms: a new method for the robust estimation of the calibration matrix.

The paper provides a new technique based on a least-squares approach for the accurate estimation of a force platform calibration matrix using simple manual procedures, when the direction of the applied loads cannot be perfectly aligned with the axes of the platform. This new procedure can be applied to all force platforms and allows the combined application of vertical and horizontal forces, both static and time-varying. The robust calibration method includes the angular errors in the least-squares parameter vector, thus reducing the bias in the estimated calibration matrix parameters. The performance of the robust method was compared with the conventional one, using a numerical simulation approach starting from a known calibration matrix. With the conventional approach, in noiseless conditions, the maximum error due to load misalignment (SD = 3 degrees) was 6% for the direct terms and over 10% for the cross-talk terms. With the robust method, these errors reduced to zero and were always below 0.4%, even when realistic noise was superimposed on the measures. With perfectly aligned loads and realistic output noise, the confidence intervals of the calibration matrix parameters were very similar for the two methods, demonstrating that the increased number of parameters did not affect the reliability of the estimate.