Torque and cadence tracking in functional electrical stimulation induced cycling using passivity-based spatial repetitive learning control

• Published in: Automatica (Oxford) Vol. 115; p. 108852
• Main Authors: Duenas, Victor H., Cousin, Christian A., Ghanbari, Vahideh, Fox, Emily J., Dixon, Warren E.
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2020
• Subjects: FES-cycling; Functional Electrical Stimulation (FES); Human–machine interaction; Passivity-based control; Spatial repetitive learning control (RLC); FES-cycling; Passivity-based control; Functional Electrical Stimulation (FES); Human–machine interaction; Spatial repetitive learning control (RLC)
• ISSN: 0005-1098; 1873-2836
• DOI: 10.1016/j.automatica.2020.108852

Due to the inherent periodic nature of cycling tasks, iterative and repetitive learning controllers are well motivated for rehabilitative cycling. Motorized functional electrical stimulation induced cycling is a rehabilitation treatment where multiple lower-limb muscle groups are activated jointly with an electric motor to achieve cycling objectives such as speed (cadence) and torque tracking. This paper examines torque tracking accomplished by the stimulation of six lower-limb muscles via a novel spatial repetitive learning control and cadence regulation by an electric motor using a sliding-mode controller. A desired torque trajectory is constructed based on the rider’s kinematic efficiency, which is a function of the crank position. The learning controller takes advantage of the periodicity of the desired torque trajectory to provide a feedforward input to the stimulated muscles. A passivity-based analysis is developed to ensure stability of the torque and cadence closed-loop error systems. The muscle learning and electric motor controllers were implemented in real-time during cycling experiments on five able-bodied individuals and three participants with movement disorders. The combined average cadence tracking error was 0.01±1.20 RPM for a 50 RPM trajectory and the combined average power tracking error was 1.78±1.25 W for a peak power of 10 W.