INERTIAL GAIT PHASE DETECTION: POLYNOMIAL NULLSPACE APPROACH

Detection of gait phases (swing and stance or more detailed) is a usual mean of feedback in closed-loop control of Functional Electrical Stimulation (FES) used for improving gait of persons with some motor control disability in a lower extremity, e.g., drop-foot problem. This paper presents a use of linear fault detection technique based on polynomial matrices for detection of gait phases. Data collected from two-axial accelerometer placed on a shank is lead to detector producing a residual signal which indicates an event by getting away from zero. The residual generator relies on linearised model of accelerometer measurement during stance phase to detect rising heel. Detection of heel rise together with threshold detection of peaks in measured signals during heel-strike and several empiric constraints (e.g., timeouts) forms the gait phase detector.