A Real-time Approach to Nonlinear Modeling of Inertial Errors with Application to Three Dimension Vehicle Navigation

Z. Shen, M.J. Korenberg, A. Noureldin

Abstract:	This paper discusses a real-time embedded solution to low cost land vehicular navigation using the digital signal processor (DSP). A reduced inertial sensor system (RISS) is utilized in this study to ensure continuous vehicular navigation. It consists of a single-axis Micro-Electro- Mechanical-System (MEMS) grade gyroscope, two MEMS grade single-axis accelerometers, and the vehicle built-in odometer. RISS is employed along with GPS to provide 3D navigation in a loosely couple scheme. Although the commonly used Kalman filter (KF) offers adequate performance when integrating high-end inertial sensors with GPS, it is unable to handle the intensive nonlinear inertial errors that jeopardize the accuracy of the low cost MEMS grade RISS. To overcome the limitation of KF, a nonlinear modeling technique, Fast Orthogonal Search (FOS) algorithm, is proposed to cascade with KF to establish a KF/FOS module. This module can model, predict, and compensate both linear and nonlinear inertial errors. Thus, it is supposed to enhance the accuracy of MEMS grade RISS drastically during GPS outages. Due to its strong computing capacity, Texas Instrument's (TI) DSP TMS320DM642 is selected as the embedded platform to implement KF/FOS module. The purpose is to apply real-time 3D RISS/GPS data fusion in terms of the proposed method so that the whole design can enable the online navigation service. The suggested approach is examined by road test trajectory data. Its online navigation outputs are compared to KFonly solution with respect to the reference. The experimental results demonstrate the advantages of the proposed method on the real-time performance and the accuracy enhancement.
Published in:	Proceedings of the 2011 International Technical Meeting of The Institute of Navigation January 24 - 26, 2011 Catamaran Resort Hotel San Diego, CA
Pages:	811 - 818
Cite this article:	Shen, Z., Korenberg, M.J., Noureldin, A., "A Real-time Approach to Nonlinear Modeling of Inertial Errors with Application to Three Dimension Vehicle Navigation," Proceedings of the 2011 International Technical Meeting of The Institute of Navigation, San Diego, CA, January 2011, pp. 811-818.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In