Quantitative comparison of time-varying system identification methods to describe human joint impedance

Accurate and swift tuning of joint impedance is crucial to perform movement interaction with our environment. Time-varying system identification enables quantification during movement. Many methods have been developed over the years, each their own mathematical approach underlying assumptions. Yet, for impedance, a systematic comparison revealing method's unique strengths weaknesses, lacking. Here, we propose quantitative framework compare these methods. The used review five time-varying using both simulated data experimental data. These included three time-domain methods: ensemble, short segment, basis impulse response function; two frequency-domain ensemble spectral, kernel-based regression. In simulation study, stiffness – static component was as square wave mimic most extreme case behavior. results were compared based on (1) variance accounted (VAF), (2) bias, (3) random, (4) total estimation error respect stiffness; (5) rise time between levels. human ankle identified. Identification performance variability in estimating representative random VAF. metrics revealed distinct properties method. Therefore, researchers must make well-justified decision which method appropriate application. combination work extensive creates assessment newly