In vision-aided INS/GNSS integrated navigation systems, several errors contribute to significant drifts in the absence of GNSS and during severe GNSS interruptions. These errors includes cameras distortion, camera calibration, camera misalignments, and feature extraction/tracking errors. In addition, inertial sensors biases, drifts, non- orthogonality, non-normality and misalignment also significantly affect the dead-reckoning navigation accuracy in the absence of GNSS. Due to the complex nonlinearity inherent in these errors, this paper introduces a double-filter mechanism enhanced by nonlinear error modeling for multi-sensor vision-aided integrated navigation systems for land vehicles. The proposed system integrates GNSS, vision, INS, and odometry. The proposed methodology tries to reduce the effects of nonlinear errors that propagate through different components of the integrated navigation system. The proposed methodology can work in real-time due to the utilization of a fast orthogonal search (FOS) approach as the nonlinear modeling technique. The paper verifies the operational performance of the proposed system using a physical experimental setup and real-road experimental data. Initial results shows remarkable positional and heading accuracy improvements if FOS training is properly performed.

Additional Metadata
Conference 27th International Technical Meeting of the Satellite Division of the Institute of Navigation, ION GNSS 2014
Citation
Nematallah, H. (H.), Korenberg, M. (M.), Noureldin, A. (A.), & Atia, M. (2014). Enhancing vision-aided GNSS/INS navigation systems using nonlinear modeling techniques based on fast orthogonal search with double-filtering mechanism. In 27th International Technical Meeting of the Satellite Division of the Institute of Navigation, ION GNSS 2014 (pp. 627–634).