Functional electrical stimulation (FES) neuroprostheses have long been a feature in the rehabilitation and gait support of stroke survivors or people with a spinal cord injury (SCI). The established FES neuroprostheses trigger the stimulation by gait events, but have to be manually tuned and cannot adapt to the persons’ individual needs. In recent research, iterative learning controllers (ILC) were developed to adapt a full stimulation pattern during the swing phase by measuring previous steps’ joint angles. However, these current approaches require manual customization of the reference joint angle trajectories to the specific subject and situation, and support only single or synergetic muscle groups. Another issue is that the measurement of joint angles with inertial measurement units (IMUs) usually involves calibration or alignment procedures, requiring time and expert knowledge. This thesis aims to develop new methods for the automatic calibration of IMU-based joint angle measurements and a learning full-leg supporting FES neuroprosthesis. Two methods for the automatic calibration of joint axes are presented: the first method can identify the joint axes of two-degrees-of-freedom (DOF) joints from arbitrary motion, and the second method can identify three joint axes of the hip and knee joint from gait motion. An ILC-based FES neuroprosthesis is developed, which consists of several new methods. This neuroprosthesis can control the antagonistic muscle pairs for knee flexion and extension and the ankle joint dorsi and plantar flexion during all gait phases. A new method for a subject and situation-specific generation of joint angle references is presented. With the introduction of physiological joint angle ranges, this method can assess each gait cycle in a meaningful way without manual tuning. An ILC was extended to operate in the gait cycle percentage domain, allowing the controller to be independent of the estimated cycle duration. The proposed 2-DOF calibration method was evaluated in an optical gait laboratory with one non‐disabled subject. It converged in under 9.5 s on average, and the root mean square (RMS) errors with respect to the optical reference system were 2.7° for the flexion/extension and 3.8° for the pronation/supination angle. The proposed 3-DOF gaitbased method was evaluated on a treadmill with eight non-disabled subjects and converged onto the expected axes in less than 15 s on average. Additionally, an existing IMU-based gait phase detection method was evaluated with available data from an optical gait laboratory. In a first test with four people with an incomplete SCI, the results showed that the proposed neuroprosthesis could generate individually fitted stimulation patterns for three of the participants. Improvements in the range of 4° at times of peak dorsiflexion, peak plantar flexion, or peak knee flexion could be measured for two subjects. The presented methods aim to push adaptive neuroprostheses further toward clinical application by reducing both the setup time and the complexity: The sensors do not have to be carefully aligned and can calibrate automatically; the neuroprosthesis’s automatic tuning procedure averaged to a duration of 139 s. Due to the customized reference joint angle generation and the introduction of physiological joint angle ranges, the subjects’ performance can be assessed automatically for different gait strategies, pathologies, and gait speeds. The resulting assist-as-needed stimulation and sensory feedback could potentially further enhance the rehabilitation effects.


    Zugriff

    Download

    Verfügbarkeit in meiner Bibliothek prüfen


    Exportieren, teilen und zitieren



    Titel :

    Learning multichannel FES gait support and inertial-sensor-based motion assessment


    Weitere Titelangaben:

    Lernende Mehrkanal-FES-Gangunterstützung und Intertialsensor-basierte Bewegungsanalyse


    Beteiligte:
    Müller, Philipp (Autor:in) / Technische Universität Berlin (Gastgebende Institution)

    Erscheinungsdatum :

    2022



    Medientyp :

    Sonstige


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Klassifikation :

    DDC:    629 / 610



    Inertial sensor-based gait recognition

    Šprager, Sebastijan / Jurič, Matjaž B. | BASE | 2021

    Freier Zugriff

    GAIT Based Inertial Navigation System Using Mobile Sensor

    Tharan, S. Sakthi / Vigneshwar, M. Vishnu / Srikumar, R. | IEEE | 2017


    An Inertial Sensor-Based Gait Analysis Pipeline for the Assessment of Real-World Stair Ambulation Parameters

    Roth, Nils / Küderle, Arne / Prossel, Domink et al. | BASE | 2021

    Freier Zugriff

    Assessment of Gait Symmetry and Gait Normality Using Inertial Sensors : In-Lab and In-Situ Evaluation

    Sant'Anna, Anita / Wickström, Nicholas / Eklund, Helene et al. | BASE | 2013

    Freier Zugriff

    Detection of Gait From Continuous Inertial Sensor Data Using Harmonic Frequencies

    Ullrich, Martin / Küderle, Arne / Hannink, Julius et al. | BASE | 2020

    Freier Zugriff