Real time estimation of time-varying human multijoint arm viscoelasticity during movements

Abstract

The mechanical impedance of the human arm has an essential role in dexterous arm control. We have developed a real time estimation method of time-varying human multijoint arm viscoelasticity during movements. By using pre-estimated arm structural parameters which are independent of arm posture, the number of estimates was reduced. To extract positional shift caused by a pseudo-random perturbation during movement, a linear band-pass filter was applied. Time-varying multijoint viscoelastic parameters were then estimated by using a recursive least-square algorithm with a forgetting factor. This method enable us to observe viscoelastic change in a single trial, and to offer the real time feedback of viscoelasticity to a subject.