Wearable pressure sensing for intelligent gesture recognition

Liu, Yuchi (2023) Wearable pressure sensing for intelligent gesture recognition. PhD thesis, University of Glasgow.

Abstract

The development of wearable sensors has become a major area of interest due to their wide range of promising applications, including health monitoring, human motion detection, human-machine interfaces, electronic skin and soft robotics. Particularly, pressure sensors have attracted considerable attention in wearable applications. However, traditional pressure sensing systems are using rigid sensors to detect the human motions. Lightweight and flexible pressure sensors are required to improve the comfortability of devices. Furthermore, in comparison with conventional sensing techniques without smart algorithm, machine learning-assisted wearable systems are capable of intelligently analysing data for classification or prediction purposes, making the system ‘smarter’ for more demanding tasks. Therefore, combining flexible pressure sensors and machine learning is a promising method to deal with human motion recognition.

This thesis focuses on fabricating flexible pressure sensors and developing wearable applications to recognize human gestures. Firstly, a comprehensive literature review was conducted, including current state-of-the-art on pressure sensing techniques and machine learning algorithms. Secondly, a piezoelectric smart wristband was developed to distinguish finger typing movements. Three machine learning algorithms, K Nearest Neighbour (KNN), Decision Tree (DT) and Support Vector Machine (SVM), were used to classify the movement of different fingers. The SVM algorithm outperformed other classifiers with an overall accuracy of 98.67% and 100% when processing raw data and extracted features.

Thirdly, a piezoresistive wristband was fabricated based on a flake-sphere composite configuration in which reduced graphene oxide fragments are doped with polystyrene spheres to achieve both high sensitivity and flexibility. The flexible wristband measured the pressure distribution around the wrist for accurate and comfortable hand gesture classification. The intelligent wristband was able to classify 12 hand gestures with 96.33% accuracy for five participants using a machine learning algorithm. Moreover, for demonstrating the practical applications of the proposed method, a realtime system was developed to control a robotic hand according to the classification results.

Finally, this thesis also demonstrates an intelligent piezoresistive sensor to recognize different throat movements during pronunciation. The piezoresistive sensor was fabricated using two PolyDimethylsiloxane (PDMS) layers that were coated with silver nanowires and reduced graphene oxide films, where the microstructures were fabricated by the polystyrene spheres between the layers. The highly sensitive sensor was able to distinguish throat vibrations from five different spoken words with an accuracy of 96% using the artificial neural network algorithm.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: T Technology > T Technology (General); T Technology > TA Engineering (General). Civil engineering (General)
• Colleges/Schools: College of Science and Engineering > School of Engineering
• Supervisor's Name: Ghannam, Dr. Rami
• Date of Award: 2023
• Depositing User: Theses Team
• Unique ID: glathesis:2023-83615
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 30 May 2023 14:10
• Last Modified: 30 May 2023 14:10
• Thesis DOI: 10.5525/gla.thesis.83615
• URI: https://theses.gla.ac.uk/id/eprint/83615

Related URLs

• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI
• Article DOI