Deep learning method for IMU-based tracking of Martian rover
Fragment książki (Rozdział monografii pokonferencyjnej)
MNiSW
5
spoza wykazu
| Status: | |
| Autorzy: | Tomiło Paweł |
| Dyscypliny: | |
| Aby zobaczyć szczegóły należy się zalogować. | |
| Wersja dokumentu: | Drukowana | Elektroniczna |
| Język: | angielski |
| Strony: | 2 - 6 |
| Efekt badań statutowych | NIE |
| Materiał konferencyjny: | TAK |
| Nazwa konferencji: | 16th European-African Regional Conference of the ISTVS |
| Skrócona nazwa konferencji: | 16th EARC ISTVS 2023 |
| URL serii konferencji: | LINK |
| Termin konferencji: | 11 października 2023 do 13 października 2023 |
| Miasto konferencji: | Lublin |
| Państwo konferencji: | POLSKA |
| Publikacja OA: | NIE |
| Abstrakty: | angielski |
| The double integration method is one of the most used techniques for calculating an object's position in time. As dynamic noise and bias can affect sensor data, filters like the alpha-beta filter or the Kalman filter are frequently utilized. They are in capable of compensating for sensor reading noise. The quality of the sensor has a big impact on how accurate these solutions are. When using inexpensive sensors, a correctly built neural network model can help compensate for noise and determinate the change in the position of the object over time. The article describes the method of tracking an RC vehicle using the low-cost IMU module and software consisting of an artificial neural network model specially developed for this purpose. Base stations that emitted infrared laser scans were utilized as a reference method to confirm the accuracy of the proposed solution and as a means of data collection for training. The proposed model provides acceptable accuracy when compared to the reference method. the tests were carried out on a short section of the road, in the case of longer sections the estimation error may be larger. The developed solution can be used as part of a larger system combined with visual odometry to navigate an autonomous vehicle. |
