As initial step I am now working on calibration of the "force" part of the calibration matrix. This will drastically semplify the data collection. This will also permit to calibrate part of the identifiable inertial parameters.
This two calibration step will then be validated by verifying that estimated external forces (not torques) in different configuration of the leg without any external contact are appropriately small. (Currently we get mismatches in the order of tens of Newtons).
As initial step I am now working on calibration of the "force" part of the calibration matrix. This will drastically semplify the data collection. This will also permit to calibrate part of the identifiable inertial parameters. This two calibration step will then be validated by verifying that estimated external forces (not torques) in different configuration of the leg without any external contact are appropriately small. (Currently we get mismatches in the order of tens of Newtons).