Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors
The estimation of the speed of human motion from wearable IMU sensors is required in applications such as pedestrian dead reckoning. In this paper, we test deep learning methods for the prediction of the motion speed from raw readings of a low-cost IMU sensor. Each subject was observed using three s...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-05-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/22/10/3865 |
| _version_ | 1827666546557714432 |
|---|---|
| author | Josef Justa Václav Šmídl Aleš Hamáček |
| author_facet | Josef Justa Václav Šmídl Aleš Hamáček |
| author_sort | Josef Justa |
| collection | DOAJ |
| description | The estimation of the speed of human motion from wearable IMU sensors is required in applications such as pedestrian dead reckoning. In this paper, we test deep learning methods for the prediction of the motion speed from raw readings of a low-cost IMU sensor. Each subject was observed using three sensors at the shoe, shin, and thigh. We show that existing general-purpose architectures outperform classical feature-based approaches and propose a novel architecture tailored for this task. The proposed architecture is based on a semi-supervised variational auto-encoder structure with innovated decoder in the form of a dense layer with a sinusoidal activation function. The proposed architecture achieved the lowest average error on the test data. Analysis of sensor placement reveals that the best location for the sensor is the shoe. Significant accuracy gain was observed when all three sensors were available. All data acquired in this experiment and the code of the estimation methods are available for download. |
| first_indexed | 2024-03-10T01:53:11Z |
| format | Article |
| id | doaj.art-a475305f582c4e8ea92666783f2d4813 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T01:53:11Z |
| publishDate | 2022-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-a475305f582c4e8ea92666783f2d48132023-11-23T13:02:35ZengMDPI AGSensors1424-82202022-05-012210386510.3390/s22103865Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU SensorsJosef Justa0Václav Šmídl1Aleš Hamáček2Department of Measurement and Technology, Faculty of Electrical Engineering, University of West Bohemia, 30100 Pilsen, Czech RepublicReseach and Innovation Center, Faculty of Electrical Engineering, University of West Bohemia, 30100 Pilsen, Czech RepublicDepartment of Measurement and Technology, Faculty of Electrical Engineering, University of West Bohemia, 30100 Pilsen, Czech RepublicThe estimation of the speed of human motion from wearable IMU sensors is required in applications such as pedestrian dead reckoning. In this paper, we test deep learning methods for the prediction of the motion speed from raw readings of a low-cost IMU sensor. Each subject was observed using three sensors at the shoe, shin, and thigh. We show that existing general-purpose architectures outperform classical feature-based approaches and propose a novel architecture tailored for this task. The proposed architecture is based on a semi-supervised variational auto-encoder structure with innovated decoder in the form of a dense layer with a sinusoidal activation function. The proposed architecture achieved the lowest average error on the test data. Analysis of sensor placement reveals that the best location for the sensor is the shoe. Significant accuracy gain was observed when all three sensors were available. All data acquired in this experiment and the code of the estimation methods are available for download.https://www.mdpi.com/1424-8220/22/10/3865motion speed estimationinertial measurement unitdeep learningwalking speedautoencoder architecture |
| spellingShingle | Josef Justa Václav Šmídl Aleš Hamáček Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors Sensors motion speed estimation inertial measurement unit deep learning walking speed autoencoder architecture |
| title | Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors |
| title_full | Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors |
| title_fullStr | Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors |
| title_full_unstemmed | Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors |
| title_short | Deep Learning Methods for Speed Estimation of Bipedal Motion from Wearable IMU Sensors |
| title_sort | deep learning methods for speed estimation of bipedal motion from wearable imu sensors |
| topic | motion speed estimation inertial measurement unit deep learning walking speed autoencoder architecture |
| url | https://www.mdpi.com/1424-8220/22/10/3865 |
| work_keys_str_mv | AT josefjusta deeplearningmethodsforspeedestimationofbipedalmotionfromwearableimusensors AT vaclavsmidl deeplearningmethodsforspeedestimationofbipedalmotionfromwearableimusensors AT aleshamacek deeplearningmethodsforspeedestimationofbipedalmotionfromwearableimusensors |