User intention detection algorithm for a lower limb exoskeleton using kinematics and kinetics data

Lower limb exoskeletons for rehabilitation purposes can help address the need for rehabilitation within victims of lower limb mobility impairments. To provide effective control over the motion of the exoskeleton, an accurate gait phase detection system must be utilized to identify the intent of the...

Full description

Saved in:
Bibliographic Details
Main Author: Boey, Chung Yin
Other Authors: Ang Wei Tech
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2024
Subjects:
Online Access:https://hdl.handle.net/10356/176644
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Lower limb exoskeletons for rehabilitation purposes can help address the need for rehabilitation within victims of lower limb mobility impairments. To provide effective control over the motion of the exoskeleton, an accurate gait phase detection system must be utilized to identify the intent of the user and provide the necessary assistance to guide the user along the intended trajectory. The Hidden Markov Model (HMM) is a commonly used machine learning method to predict the gait phase of the user based on sensor data. In this experiment, Inertial Measurement Units are used to measure lower limb angles, angular velocity, and angular acceleration in the sagittal plane, and Force Sensitive Resistors are used to determine contact points of the user’s feet. Data from 4 subjects were collected and labelled using a threshold-based algorithm to train an HMM. Subjects were tasked to perform various walking activities on level ground, obstacle crossing, inclined surfaces, and staircase. Results from the HMM were validated with the labelled data. 4 gait phases were detected by the algorithm, Foot Flat (FF), Heel Off (HO), Swing (SW), and Initial Contact (IC). Results show possible omission of the FSR and only using IMUs for an HMM algorithm to predict gait phases for a user wearing an exoskeleton in real-time.