Deep Gait Tracking with Inertial Measurement Unit

Jien De Sui; Tian Sheuan Chang

doi:10.1109/LSENS.2019.2947625

Deep Gait Tracking with Inertial Measurement Unit

Jien De Sui^*, Tian Sheuan Chang

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

3 Scopus citations

Abstract

This letter presents a convolutional neural network based foot motion tracking with only six-axis inertial-measurement-unit (IMU) sensor data. The presented approach can adapt to various walking conditions by adopting differential and window based input. The training data are further augmented by sliding and random window samplings on IMU sensor data to increase data diversity for better performance. The proposed approach fuses predictions of three-dimensional output into one model. The proposed fused model can achieve average error of 2.30 ± 2.23 cm in the X-axis, 0.91 ± 0.95 cm in the Y-axis, and 0.58 ± 0.52 cm in the Z-axis.

Original language	English
Article number	8871330
Journal	IEEE Sensors Letters
Volume	3
Issue number	11
DOIs	https://doi.org/10.1109/LSENS.2019.2947625
State	Published - Nov 2019

Keywords

convolutional neural networks (CNNs)
gait analysis
gait parameter
IMU sensor
inertial-measurement-unit (IMU) sensor
Sensor signal processing

Access to Document

10.1109/LSENS.2019.2947625

Cite this

@article{134bf01d1b6d4382aef853fa9c8249a2,

title = "Deep Gait Tracking with Inertial Measurement Unit",

abstract = "This letter presents a convolutional neural network based foot motion tracking with only six-axis inertial-measurement-unit (IMU) sensor data. The presented approach can adapt to various walking conditions by adopting differential and window based input. The training data are further augmented by sliding and random window samplings on IMU sensor data to increase data diversity for better performance. The proposed approach fuses predictions of three-dimensional output into one model. The proposed fused model can achieve average error of 2.30 ± 2.23 cm in the X-axis, 0.91 ± 0.95 cm in the Y-axis, and 0.58 ± 0.52 cm in the Z-axis.",

keywords = "convolutional neural networks (CNNs), gait analysis, gait parameter, IMU sensor, inertial-measurement-unit (IMU) sensor, Sensor signal processing",

author = "Sui, {Jien De} and Chang, {Tian Sheuan}",

year = "2019",

month = nov,

doi = "10.1109/LSENS.2019.2947625",

language = "English",

volume = "3",

journal = "IEEE Sensors Letters",

issn = "2475-1472",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "11",

}

TY - JOUR

T1 - Deep Gait Tracking with Inertial Measurement Unit

AU - Sui, Jien De

AU - Chang, Tian Sheuan

PY - 2019/11

Y1 - 2019/11

N2 - This letter presents a convolutional neural network based foot motion tracking with only six-axis inertial-measurement-unit (IMU) sensor data. The presented approach can adapt to various walking conditions by adopting differential and window based input. The training data are further augmented by sliding and random window samplings on IMU sensor data to increase data diversity for better performance. The proposed approach fuses predictions of three-dimensional output into one model. The proposed fused model can achieve average error of 2.30 ± 2.23 cm in the X-axis, 0.91 ± 0.95 cm in the Y-axis, and 0.58 ± 0.52 cm in the Z-axis.

AB - This letter presents a convolutional neural network based foot motion tracking with only six-axis inertial-measurement-unit (IMU) sensor data. The presented approach can adapt to various walking conditions by adopting differential and window based input. The training data are further augmented by sliding and random window samplings on IMU sensor data to increase data diversity for better performance. The proposed approach fuses predictions of three-dimensional output into one model. The proposed fused model can achieve average error of 2.30 ± 2.23 cm in the X-axis, 0.91 ± 0.95 cm in the Y-axis, and 0.58 ± 0.52 cm in the Z-axis.

KW - convolutional neural networks (CNNs)

KW - gait analysis

KW - gait parameter

KW - IMU sensor

KW - inertial-measurement-unit (IMU) sensor

KW - Sensor signal processing

UR - http://www.scopus.com/inward/record.url?scp=85082633914&partnerID=8YFLogxK

U2 - 10.1109/LSENS.2019.2947625

DO - 10.1109/LSENS.2019.2947625

M3 - Article

AN - SCOPUS:85082633914

SN - 2475-1472

VL - 3

JO - IEEE Sensors Letters

JF - IEEE Sensors Letters

IS - 11

M1 - 8871330

ER -

Deep Gait Tracking with Inertial Measurement Unit

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this