RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters
Inertial-sensor-based attitude estimation is a crucial technology in various applications, from human motion tracking to autonomous aerial and ground vehicles. Application scenarios differ in characteristics of the performed motion, presence of disturbances, and environmental conditions. Since state...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-09-01
|
| Series: | AI |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2673-2688/2/3/28 |
| _version_ | 1797520528338059264 |
|---|---|
| author | Daniel Weber Clemens Gühmann Thomas Seel |
| author_facet | Daniel Weber Clemens Gühmann Thomas Seel |
| author_sort | Daniel Weber |
| collection | DOAJ |
| description | Inertial-sensor-based attitude estimation is a crucial technology in various applications, from human motion tracking to autonomous aerial and ground vehicles. Application scenarios differ in characteristics of the performed motion, presence of disturbances, and environmental conditions. Since state-of-the-art attitude estimators do not generalize well over these characteristics, their parameters must be tuned for the individual motion characteristics and circumstances. We propose <i>RIANN</i>, a ready-to-use, neural network-based, parameter-free, real-time-capable inertial attitude estimator, which generalizes well across different motion dynamics, environments, and sampling rates, without the need for application-specific adaptations. We gather six publicly available datasets of which we exploit two datasets for the method development and the training, and we use four datasets for evaluation of the trained estimator in three different test scenarios with varying practical relevance. Results show that <i>RIANN</i> outperforms state-of-the-art attitude estimation filters in the sense that it generalizes much better across a variety of motions and conditions in different applications, with different sensor hardware and different sampling frequencies. This is true even if the filters are tuned on each individual test dataset, whereas <i>RIANN</i> was trained on completely separate data and has never seen any of these test datasets. <i>RIANN</i> can be applied directly without adaptations or training and is therefore expected to enable plug-and-play solutions in numerous applications, especially when accuracy is crucial but no ground-truth data is available for tuning or when motion and disturbance characteristics are uncertain. We made <i>RIANN</i> publicly available. |
| first_indexed | 2024-03-10T07:58:04Z |
| format | Article |
| id | doaj.art-6d42a2de03bb4c5086d03546523661b8 |
| institution | Directory Open Access Journal |
| issn | 2673-2688 |
| language | English |
| last_indexed | 2024-03-10T07:58:04Z |
| publishDate | 2021-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | AI |
| spelling | doaj.art-6d42a2de03bb4c5086d03546523661b82023-11-22T11:40:09ZengMDPI AGAI2673-26882021-09-012344446310.3390/ai2030028RIANN—A Robust Neural Network Outperforms Attitude Estimation FiltersDaniel Weber0Clemens Gühmann1Thomas Seel2Electronic Measurement and Diagnostic Technology, Technische Universität Berlin, 10587 Berlin, GermanyElectronic Measurement and Diagnostic Technology, Technische Universität Berlin, 10587 Berlin, GermanyDepartment Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universität Erlangen-Nürnberg, 91052 Erlangen, GermanyInertial-sensor-based attitude estimation is a crucial technology in various applications, from human motion tracking to autonomous aerial and ground vehicles. Application scenarios differ in characteristics of the performed motion, presence of disturbances, and environmental conditions. Since state-of-the-art attitude estimators do not generalize well over these characteristics, their parameters must be tuned for the individual motion characteristics and circumstances. We propose <i>RIANN</i>, a ready-to-use, neural network-based, parameter-free, real-time-capable inertial attitude estimator, which generalizes well across different motion dynamics, environments, and sampling rates, without the need for application-specific adaptations. We gather six publicly available datasets of which we exploit two datasets for the method development and the training, and we use four datasets for evaluation of the trained estimator in three different test scenarios with varying practical relevance. Results show that <i>RIANN</i> outperforms state-of-the-art attitude estimation filters in the sense that it generalizes much better across a variety of motions and conditions in different applications, with different sensor hardware and different sampling frequencies. This is true even if the filters are tuned on each individual test dataset, whereas <i>RIANN</i> was trained on completely separate data and has never seen any of these test datasets. <i>RIANN</i> can be applied directly without adaptations or training and is therefore expected to enable plug-and-play solutions in numerous applications, especially when accuracy is crucial but no ground-truth data is available for tuning or when motion and disturbance characteristics are uncertain. We made <i>RIANN</i> publicly available.https://www.mdpi.com/2673-2688/2/3/28attitude estimationnonlinear filtersinertial sensorsinformation fusionneural networksrecurrent neural networks |
| spellingShingle | Daniel Weber Clemens Gühmann Thomas Seel RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters AI attitude estimation nonlinear filters inertial sensors information fusion neural networks recurrent neural networks |
| title | RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters |
| title_full | RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters |
| title_fullStr | RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters |
| title_full_unstemmed | RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters |
| title_short | RIANN—A Robust Neural Network Outperforms Attitude Estimation Filters |
| title_sort | riann a robust neural network outperforms attitude estimation filters |
| topic | attitude estimation nonlinear filters inertial sensors information fusion neural networks recurrent neural networks |
| url | https://www.mdpi.com/2673-2688/2/3/28 |
| work_keys_str_mv | AT danielweber riannarobustneuralnetworkoutperformsattitudeestimationfilters AT clemensguhmann riannarobustneuralnetworkoutperformsattitudeestimationfilters AT thomasseel riannarobustneuralnetworkoutperformsattitudeestimationfilters |