Accuracy of human joint impedance identification

Human joint impedance represents the dynamical relationship between the torque and the angle of a joint of the human muscoskeletal system. The identification of human joint impedance of the lower limbs’ joints can provide important information for the design and control of wearable bionic devices that can assist the user in performing daily tasks. However, there is only limited research on the identification of joint impedance during such tasks, where the joint impedance should be modeled by a time-varying system.

In this study, a consistent estimator is proposed for the identification of human joint impedance during locomotion. Human joint impedance is modeled as a mass-spring-damper system with time-varying parameters, which are represented as the sum of sigmoidal basis functions. A realistic ankle impedance simulation is obtained to perform a Monte Carlo analysis of the proposed estimator, in order to determine the required experimental conditions (persistency of excitation, measurement time / frequency resolution) to obtain a sufficiently low uncertainty for the application at hand.
The results show that the proposed estimator can reconstruct the parameters of the system with high accuracy and low uncertainty.