Iterative learning control for correction of drop- foot using bio-impedance as sensory information

Automatic control of ankle-joint angle for the correction of drop-foot has been investigated using Bio-Impedance (BI) for measuring the joint angle and Iterative Learning Control (ILC) for adjusting the stimulation profile. A customised four-channel measurement system (50 kHz) was used to assess bio-impedance changes caused by ankle-joint motion. Two current excitation electrodes were placed below the patella on the shank and on the dorsum of the foot. Voltage detection electrodes were attached to the posterior surface of lower leg below the calf and on the M. tibialis anterior close to its origin. A demodulation circuit determines changes in the absolute value of the BI from the measured and amplified voltage. All circuits are protected against stimulation artefacts so that the recording of BI is possible while muscle stimulation is active. Calibration of the angle measurement was performed by positioning the ankle joint at three known angles. Reference measurements were taken with a marker-based optical system. Dorsiflexion of the unconstrained ankle joint was achieved by stimulation of the M. tibialis anterior. First-order ILC was applied to realise a pre-defined angle profile in a cycle-to-cycle manner. The new stimulation intensity profile will be an update of the last profile taking tracking errors of the last cycle into account. Preliminary experiments were conducted with one able-bodied subject. An almost linear correlation between ankle-joint angle and bio-impedance was found for the angle range applicable during gait. The chosen angle trajectory (sine half-wave from the resting foot to 0 degree) was realised by the ILC within 3 cycles. The final root mean square tracking error was below 5 degree. Automatic control of ankle-joint angle by ILC is feasible when using bio-impedance as sensory information. Experiments under real walking conditions and with stroke patients must be conducted in future