IMU-based deep neural networks: Prediction of locomotion and transition intentions for an osseointegrated transfemoral amputee

Bruinsma, Julian (2021) IMU-based deep neural networks: Prediction of locomotion and transition intentions for an osseointegrated transfemoral amputee. Bachelor's Thesis, Artificial Intelligence.

Preview	Text BachelorThesisFinalVersionJulianBruinsma.pdf Download (422kB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (118kB)

Abstract

This paper focuses on the design and comparison of different deep neural networks for the real-time locomotor intention prediction of one osseointegrated amputee by using data from an inertial measurement unit (IMU). The deep neural networks are based on convolutional neural networks, recurrent neural networks, and convolutional recurrent neural networks. The input to the architectures are features in both time-domain and time-frequency domain, which are derived from either one IMU placed on the upper left thigh or two IMUs placed on both the left thigh and left shank of the osseointegrated amputee. The prediction of eight and seven different locomotion modes and twenty-four and twenty transitions are investigated with or without sitting, respectively. The study shows that a recurrent network, realized with four layers of gated recurrent unit networks, achieves, with 5-fold cross-validation, a mean F1-score of 84.77% and 86.5% using one IMU and 93.06% and 89.99% using two IMUs, with or without sitting, respectively.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Carloni, R.
Degree programme:	Artificial Intelligence
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	15 Apr 2021 07:35
Last Modified:	10 Jun 2021 09:33
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/24264

Actions (login required)

View Item