Skip to main content


Fig. 1 | Journal of NeuroEngineering and Rehabilitation

Fig. 1

From: Voluntary control of wearable robotic exoskeletons by patients with paresis via neuromechanical modeling

Fig. 1

Schematic representation of the real-time modeling framework and its communication with the robotic exoskeleton. The whole framework is operated by a Raspberry Pi 3 single-board computer. The framework consists of five main components: a The EMG plugin collects muscle bioelectric signals from wearable active electrodes and transfers them to the EMG-driven model. b The B-spline component computes musculotendon length (Lmt) and moment arm (MA) values from joint angles collected via robotic exoskeleton sensors. c The EMG-driven model uses input EMG, Lmt and MA data to compute the resulting mechanical forces in 12 lower-extremity musculotendon units (Table 1) and joint moment about the degrees of freedom of knee flexion-extension and ankle plantar-dorsiflexion. d The offline calibration procedure identifies internal parameters of the model that vary non-linearly across individuals. These include optimal fiber length and tendon slack length, muscle maximal isometric force, and excitation-to-activation shape factors. e The exoskeleton plugin converts EMG-driven model-based joint moment estimates into exoskeleton control commands. Please refer to the Methods section for an in-depth description

Back to article page