aalto1 untyped-item.component.html
Intention detection for upper limb exoskeleton control using force myography and inertial measurement unit
Loading...
URL
Journal Title
Journal ISSN
Volume Title
School of Electrical Engineering |
Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Authors
Date
Department
Major/Subject
Mcode
Degree programme
Language
en
Pages
76
Series
Abstract
Job-related injuries pose a major risk in physically intensive occupations, often resulting in musculoskeletal disorders (MSD) triggered by repetitive lifting and demanding postures, such as overhead tasks. The increasing incidence of upper-limb injuries underscores the immediate demand for strategies to reduce the risks associated with overhead tasks and manipulating heavy objects. This thesis addresses these issues by focusing on adaptive upper-limb assistance through exoskeletons. It investigates recognizing user intention during motion and estimating payload as a foundation for providing adaptive support with upper-limb exoskeletons.
This research utilizes force myography (FMG) obtained through a force-sensitive resistor (FSR) alongside inertial measurement units (IMUs) to facilitate action recognition and classify payloads. For this study, a custom FSR band featuring a voltage divider circuit was built. The FSR band was placed on the biceps muscle, whereas the IMU was positioned beneath it. Together, these instruments collected essential information on arm movements and muscle engagement during numerous activities with varying load scenarios.
Initial feasibility assessments for payload classification during basic dynamic motions were performed using restricted data and machine learning techniques, such as Linear Discriminant Analysis (LDA), Support Vector Machines (SVM) and Long Short-Term Memory networks combined with Fully Convolutional Networks (LSTM-FCN). Afterwards, action recognition and payload classification were conducted for more intricate movements. Time series classification methods, including Derivative Dynamic Time Warping (DDTW) and LSTM-FCN for action identification, and Derivative Time Warping (DTW) and LSTM-FCN for payload classification, were applied.
This study advances the progress of intention recognition systems by demonstrating the capability of FMG and IMU-based system that can be utilized for control of upper-limb exoskeletons to provide dynamic assistance. Although FMG has been analyzed in prosthetics and hand gesture recognition, this research is one of the initial efforts to investigate its application in payload classification during dynamic movements.