A Novel Cable-Tightness-Based Control Strategy With High-Accuracy and Seamless Gait Assistance for Cable-Driven Exoskeletons

Abstract

Cable-driven exoskeletons face control challenges such as transmission latency and complex kinematics. In gait assistance, torque–position switching strategy can provide transparency and near-zero impedance, but its mode switching introduces discontinuities. Iterative-learning methods enhance tracking performance but cannot accommodate parameter-varying control needed for adaptive, environment-responsive assistance. We developed a cable-tightness-based controller with a unified control variable for high-accuracy and seamless gait assistance. The control variable, “tightness”, was defined through kinematic analysis as a bidirectional measure of the system's transmission state. The high-level controller characterizes the torque–tightness relationship online and maps the planned torque profile to the corresponding tightness profile. The low-level controller tracks the desired profiles through kinematics-based feedforward and torque–tightness hybrid feedback control. We deployed the proposed controller on a knee exoskeleton system, then conducted parameter tuning and performance evaluations during gait assistance. Experimental results demonstrated that the controller achieved high tracking accuracy, with an averaged RMS error of 0.85% peak torque; realized seamless control, reducing the maximum motor impulse by 91.9%; and sustained performance under parameter-varying assistance, reducing RMS error by up to 72.4%. Preliminary findings indicated the control-level significance of the proposed strategy in enhancing gait assistance.