LPV subspace identification of time-variant joint impedance

Goal of this study was to verify the usability of a novel linear parameter varying (LPV) identification technique for the identification of time-varying joint impedance. Quantification of neuromechanical parameters describing joint impedance improves understanding of human movement control and is of clinical relevance for diagnostics and treatment monitoring of patients suffering from movement disorders. For large rotations and/or changing activation of the muscles linear time invariant (LTI) identification techniques no longer suffice because the parameters describing the system are not constant over time. To identify joint impedance over a large working range nonlinear or time variant identification techniques are thus required. Time-variant damping and stiffness of the human wrist joint were estimated in a pilot experiment. The proposed method performed well, as shown by high VAF values indicating the suitability of LPV subspace identification for describing time-varying joint impedance.