Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system

R. Sharaf, M. Tarbouchi, A. El-Shafie, A. Noureldin

Research output: Chapter in Book/Report/Conference proceedingConference contribution

15 Citations (Scopus)

Abstract

Presently, Kalman filter (KF) is used to fuse data from both inertial navigation systems (INS) and global positioning systems (GPS) to provide position, velocity and attitude information. However, several drawbacks associated with KF like its immunity to noise, its dependency on predefined errors models, has encouraged research activates towards investigation of other integration techniques. This study proposes and discusses the real-time implementation of adaptive neuro-fuzzy inference system (ANFIS) to fuse GPS and INS data for vehicular navigation applications. The ANFIS model is designed to process the INS position component at its input and provide the corresponding INS position error at its output. This model is based on the Tagaki-Sugeno-Kang (TSK) fuzzy logic inference system. During the availability of the GPS signal, the ANFIS module processes the INS position components and the model parameters are updated towards their optimal values while minimizing the root mean square estimation error between the ANFIS output and the difference between the GPS and INS position components. The INS error provided by the ANFIS model is continuously removed from the corresponding INS position component. The proposed method is implemented for real-time applications through a data window of appropriate size that processes the INS position component and the corresponding INS error referenced to GPS position. This window slides along the INS and the GPS data over the entire navigation mission. During the prediction mode (upon losing the GPS satellite signal), the navigation system relies on the ANFIS module to predict INS errors and remove them from their corresponding INS position components. The proposed method was examined and compared to KF when applied to Ashtech Z12 GPS receiver and a navigation-grade INS (Honeywell LRF-III) that have been utilized inside a land vehicle. The system is evaluated while considering several intentionally introduced GPS outages for periods of 20 seconds. The ANFIS-based navigation system was able to provide the vehicle position with errors, which were below 2 m. The experimental results demonstrated the advantages of the proposed AI-based INS/GPS integration techniques in regards of robustness while ensuring system position accuracy in real-time.

Original languageEnglish
Title of host publicationProceedings of the Institute of Navigation, National Technical Meeting
Pages235-242
Number of pages8
Publication statusPublished - 2005
Externally publishedYes
EventInstitute of Navigation, 2005 National Technical Meeting, NTM 2005 - San Diego, CA
Duration: 24 Jan 200526 Jan 2005

Other

OtherInstitute of Navigation, 2005 National Technical Meeting, NTM 2005
CitySan Diego, CA
Period24/1/0526/1/05

Fingerprint

Inertial navigation systems
Data fusion
Fuzzy inference
Global positioning system
Kalman filters
Navigation
Electric fuses
Navigation systems
Outages
Error analysis
Fuzzy logic

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Sharaf, R., Tarbouchi, M., El-Shafie, A., & Noureldin, A. (2005). Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system. In Proceedings of the Institute of Navigation, National Technical Meeting (pp. 235-242)

Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system. / Sharaf, R.; Tarbouchi, M.; El-Shafie, A.; Noureldin, A.

Proceedings of the Institute of Navigation, National Technical Meeting. 2005. p. 235-242.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sharaf, R, Tarbouchi, M, El-Shafie, A & Noureldin, A 2005, Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system. in Proceedings of the Institute of Navigation, National Technical Meeting. pp. 235-242, Institute of Navigation, 2005 National Technical Meeting, NTM 2005, San Diego, CA, 24/1/05.
Sharaf R, Tarbouchi M, El-Shafie A, Noureldin A. Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system. In Proceedings of the Institute of Navigation, National Technical Meeting. 2005. p. 235-242
Sharaf, R. ; Tarbouchi, M. ; El-Shafie, A. ; Noureldin, A. / Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system. Proceedings of the Institute of Navigation, National Technical Meeting. 2005. pp. 235-242
@inproceedings{edff219f6ae443719f38e2ecbd895a3b,
title = "Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system",
abstract = "Presently, Kalman filter (KF) is used to fuse data from both inertial navigation systems (INS) and global positioning systems (GPS) to provide position, velocity and attitude information. However, several drawbacks associated with KF like its immunity to noise, its dependency on predefined errors models, has encouraged research activates towards investigation of other integration techniques. This study proposes and discusses the real-time implementation of adaptive neuro-fuzzy inference system (ANFIS) to fuse GPS and INS data for vehicular navigation applications. The ANFIS model is designed to process the INS position component at its input and provide the corresponding INS position error at its output. This model is based on the Tagaki-Sugeno-Kang (TSK) fuzzy logic inference system. During the availability of the GPS signal, the ANFIS module processes the INS position components and the model parameters are updated towards their optimal values while minimizing the root mean square estimation error between the ANFIS output and the difference between the GPS and INS position components. The INS error provided by the ANFIS model is continuously removed from the corresponding INS position component. The proposed method is implemented for real-time applications through a data window of appropriate size that processes the INS position component and the corresponding INS error referenced to GPS position. This window slides along the INS and the GPS data over the entire navigation mission. During the prediction mode (upon losing the GPS satellite signal), the navigation system relies on the ANFIS module to predict INS errors and remove them from their corresponding INS position components. The proposed method was examined and compared to KF when applied to Ashtech Z12 GPS receiver and a navigation-grade INS (Honeywell LRF-III) that have been utilized inside a land vehicle. The system is evaluated while considering several intentionally introduced GPS outages for periods of 20 seconds. The ANFIS-based navigation system was able to provide the vehicle position with errors, which were below 2 m. The experimental results demonstrated the advantages of the proposed AI-based INS/GPS integration techniques in regards of robustness while ensuring system position accuracy in real-time.",
author = "R. Sharaf and M. Tarbouchi and A. El-Shafie and A. Noureldin",
year = "2005",
language = "English",
pages = "235--242",
booktitle = "Proceedings of the Institute of Navigation, National Technical Meeting",

}

TY - GEN

T1 - Real-time implementation of INS/GPS data fusion utilizing adaptive neuro-fuzzy inference system

AU - Sharaf, R.

AU - Tarbouchi, M.

AU - El-Shafie, A.

AU - Noureldin, A.

PY - 2005

Y1 - 2005

N2 - Presently, Kalman filter (KF) is used to fuse data from both inertial navigation systems (INS) and global positioning systems (GPS) to provide position, velocity and attitude information. However, several drawbacks associated with KF like its immunity to noise, its dependency on predefined errors models, has encouraged research activates towards investigation of other integration techniques. This study proposes and discusses the real-time implementation of adaptive neuro-fuzzy inference system (ANFIS) to fuse GPS and INS data for vehicular navigation applications. The ANFIS model is designed to process the INS position component at its input and provide the corresponding INS position error at its output. This model is based on the Tagaki-Sugeno-Kang (TSK) fuzzy logic inference system. During the availability of the GPS signal, the ANFIS module processes the INS position components and the model parameters are updated towards their optimal values while minimizing the root mean square estimation error between the ANFIS output and the difference between the GPS and INS position components. The INS error provided by the ANFIS model is continuously removed from the corresponding INS position component. The proposed method is implemented for real-time applications through a data window of appropriate size that processes the INS position component and the corresponding INS error referenced to GPS position. This window slides along the INS and the GPS data over the entire navigation mission. During the prediction mode (upon losing the GPS satellite signal), the navigation system relies on the ANFIS module to predict INS errors and remove them from their corresponding INS position components. The proposed method was examined and compared to KF when applied to Ashtech Z12 GPS receiver and a navigation-grade INS (Honeywell LRF-III) that have been utilized inside a land vehicle. The system is evaluated while considering several intentionally introduced GPS outages for periods of 20 seconds. The ANFIS-based navigation system was able to provide the vehicle position with errors, which were below 2 m. The experimental results demonstrated the advantages of the proposed AI-based INS/GPS integration techniques in regards of robustness while ensuring system position accuracy in real-time.

AB - Presently, Kalman filter (KF) is used to fuse data from both inertial navigation systems (INS) and global positioning systems (GPS) to provide position, velocity and attitude information. However, several drawbacks associated with KF like its immunity to noise, its dependency on predefined errors models, has encouraged research activates towards investigation of other integration techniques. This study proposes and discusses the real-time implementation of adaptive neuro-fuzzy inference system (ANFIS) to fuse GPS and INS data for vehicular navigation applications. The ANFIS model is designed to process the INS position component at its input and provide the corresponding INS position error at its output. This model is based on the Tagaki-Sugeno-Kang (TSK) fuzzy logic inference system. During the availability of the GPS signal, the ANFIS module processes the INS position components and the model parameters are updated towards their optimal values while minimizing the root mean square estimation error between the ANFIS output and the difference between the GPS and INS position components. The INS error provided by the ANFIS model is continuously removed from the corresponding INS position component. The proposed method is implemented for real-time applications through a data window of appropriate size that processes the INS position component and the corresponding INS error referenced to GPS position. This window slides along the INS and the GPS data over the entire navigation mission. During the prediction mode (upon losing the GPS satellite signal), the navigation system relies on the ANFIS module to predict INS errors and remove them from their corresponding INS position components. The proposed method was examined and compared to KF when applied to Ashtech Z12 GPS receiver and a navigation-grade INS (Honeywell LRF-III) that have been utilized inside a land vehicle. The system is evaluated while considering several intentionally introduced GPS outages for periods of 20 seconds. The ANFIS-based navigation system was able to provide the vehicle position with errors, which were below 2 m. The experimental results demonstrated the advantages of the proposed AI-based INS/GPS integration techniques in regards of robustness while ensuring system position accuracy in real-time.

UR - http://www.scopus.com/inward/record.url?scp=27744550441&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=27744550441&partnerID=8YFLogxK

M3 - Conference contribution

SP - 235

EP - 242

BT - Proceedings of the Institute of Navigation, National Technical Meeting

ER -