Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks

Hassan  Wagih; Mostafa Osman; Mohammed I.  Awad; Sherif  Hammad

doi:10.1109/ITSC55140.2022.9921796

Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks

Hassan Wagih, Mostafa Osman, Mohammed I. Awad, Sherif Hammad

Team Manon Kok

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

22 Downloads (Pure)

Abstract

In this paper, an approach for reducing the drift in monocular visual odometry algorithms is proposed based on a feedforward neural network. A visual odometry algorithm computes the incremental motion of the vehicle between the successive camera frames, then integrates these increments to determine the pose of the vehicle. The proposed neural network reduces the errors in the pose estimation of the vehicle which results from the inaccuracies in features detection and matching, camera intrinsic parameters, and so on. These inaccuracies are propagated to the motion estimation of the vehicle causing larger amounts of estimation errors. The drift reducing neural network identifies such errors based on the motion of features in the successive camera frames leading to more accurate incremental motion estimates. The proposed drift reducing neural network is trained and validated using the KITTI dataset and the results show the efficacy of the proposed approach in reducing the errors in the incremental orientation estimation, thus reducing the overall error in the pose estimation.

Original language	English
Title of host publication	Proceedings of the IEEE 25th International Conference on Intelligent Transportation Systems (ITSC 2022)
Publisher	IEEE
Pages	1356-1361
ISBN (Electronic)	978-1-6654-6880-0
ISBN (Print)	978-1-6654-6881-7
DOIs	https://doi.org/10.1109/ITSC55140.2022.9921796
Publication status	Published - 2022
Event	2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC) - Macau, China Duration: 8 Oct 2022 → 12 Oct 2022 Conference number: 25th

Conference

Conference	2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)
Country/Territory	China
City	Macau
Period	8/10/22 → 12/10/22

Bibliographical note

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

Estimation error
Intelligent vehicles
Motion estimation
Feature detection
Pose estimation
Cameras
Feedforward neural networks

Access to Document

10.1109/ITSC55140.2022.9921796

Drift_Reduction_for_Monocular_Visual_Odometry_of_Intelligent_Vehicles_Using_Feedforward_Neural_NetworksFinal published version, 1.52 MB

Cite this

@inproceedings{faa36bcd1cfe40648df00b4350c56fd1,

title = "Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks",

abstract = "In this paper, an approach for reducing the drift in monocular visual odometry algorithms is proposed based on a feedforward neural network. A visual odometry algorithm computes the incremental motion of the vehicle between the successive camera frames, then integrates these increments to determine the pose of the vehicle. The proposed neural network reduces the errors in the pose estimation of the vehicle which results from the inaccuracies in features detection and matching, camera intrinsic parameters, and so on. These inaccuracies are propagated to the motion estimation of the vehicle causing larger amounts of estimation errors. The drift reducing neural network identifies such errors based on the motion of features in the successive camera frames leading to more accurate incremental motion estimates. The proposed drift reducing neural network is trained and validated using the KITTI dataset and the results show the efficacy of the proposed approach in reducing the errors in the incremental orientation estimation, thus reducing the overall error in the pose estimation.",

keywords = "Estimation error, Intelligent vehicles, Motion estimation, Feature detection, Pose estimation, Cameras, Feedforward neural networks",

author = "Hassan Wagih and Mostafa Osman and Awad, {Mohammed I.} and Sherif Hammad",

note = "Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.; 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC) ; Conference date: 08-10-2022 Through 12-10-2022",

year = "2022",

doi = "10.1109/ITSC55140.2022.9921796",

language = "English",

isbn = "978-1-6654-6881-7",

pages = "1356--1361",

booktitle = "Proceedings of the IEEE 25th International Conference on Intelligent Transportation Systems (ITSC 2022)",

publisher = "IEEE",

address = "United States",

}

Wagih, H, Osman, M, Awad, MI & Hammad, S 2022, Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks. in Proceedings of the IEEE 25th International Conference on Intelligent Transportation Systems (ITSC 2022). IEEE, pp. 1356-1361, 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC), Macau, China, 8/10/22. https://doi.org/10.1109/ITSC55140.2022.9921796

Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks. / Wagih, Hassan; Osman, Mostafa; Awad, Mohammed I. et al.
Proceedings of the IEEE 25th International Conference on Intelligent Transportation Systems (ITSC 2022). IEEE, 2022. p. 1356-1361.

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks

AU - Wagih, Hassan

AU - Osman, Mostafa

AU - Awad, Mohammed I.

AU - Hammad, Sherif

N1 - Conference code: 25th

PY - 2022

Y1 - 2022

N2 - In this paper, an approach for reducing the drift in monocular visual odometry algorithms is proposed based on a feedforward neural network. A visual odometry algorithm computes the incremental motion of the vehicle between the successive camera frames, then integrates these increments to determine the pose of the vehicle. The proposed neural network reduces the errors in the pose estimation of the vehicle which results from the inaccuracies in features detection and matching, camera intrinsic parameters, and so on. These inaccuracies are propagated to the motion estimation of the vehicle causing larger amounts of estimation errors. The drift reducing neural network identifies such errors based on the motion of features in the successive camera frames leading to more accurate incremental motion estimates. The proposed drift reducing neural network is trained and validated using the KITTI dataset and the results show the efficacy of the proposed approach in reducing the errors in the incremental orientation estimation, thus reducing the overall error in the pose estimation.

AB - In this paper, an approach for reducing the drift in monocular visual odometry algorithms is proposed based on a feedforward neural network. A visual odometry algorithm computes the incremental motion of the vehicle between the successive camera frames, then integrates these increments to determine the pose of the vehicle. The proposed neural network reduces the errors in the pose estimation of the vehicle which results from the inaccuracies in features detection and matching, camera intrinsic parameters, and so on. These inaccuracies are propagated to the motion estimation of the vehicle causing larger amounts of estimation errors. The drift reducing neural network identifies such errors based on the motion of features in the successive camera frames leading to more accurate incremental motion estimates. The proposed drift reducing neural network is trained and validated using the KITTI dataset and the results show the efficacy of the proposed approach in reducing the errors in the incremental orientation estimation, thus reducing the overall error in the pose estimation.

KW - Estimation error

KW - Intelligent vehicles

KW - Motion estimation

KW - Feature detection

KW - Pose estimation

KW - Cameras

KW - Feedforward neural networks

UR - http://www.scopus.com/inward/record.url?scp=85141817946&partnerID=8YFLogxK

U2 - 10.1109/ITSC55140.2022.9921796

DO - 10.1109/ITSC55140.2022.9921796

M3 - Conference contribution

SN - 978-1-6654-6881-7

SP - 1356

EP - 1361

BT - Proceedings of the IEEE 25th International Conference on Intelligent Transportation Systems (ITSC 2022)

PB - IEEE

T2 - 2022 IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)

Y2 - 8 October 2022 through 12 October 2022

ER -

Drift Reduction for Monocular Visual Odometry of Intelligent Vehicles Using Feedforward Neural Networks

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this