Particle Filtering to Improve GPS/INS Integration

Howard L. Dyckman, Shon Sloat and Bill Pettus

Abstract:	Navigation based upon the Global Positioning System (GPS) fused with Inertial Navigation System (INS) and other sensors has typically been performed by means of Kalman Filter (KF) implementation. More precisely, the linear model assumptions of the KF are generally approximated by linearizing the navigation model to conform to the KF scope of application. This linearized KF implementation is known as the Extended Kalman Filter (EKF). While Kalman filtering represents a generalized optimal solution for the case of a linear model and Gaussian noise assumptions, it becomes a suboptimal approach as these assumptions are violated. Consequently, alternative estimation strategies become available. In this paper the Particle Filter (PF) [1-8] is presented as an attractive alternative to the EKF for navigation applications. PF in general represents an approach whereby a succession of Monte Carlo simulations is performed in a fashion that progressively evolves towards a desired solution. Since this simulation process utilizes the full scope of system model and noise conditions, Gaussian system noise and linearization approximations are no longer necessary as conditions that determine solution performance or viability. Moreover, the very nature of a model simulation strategy greatly simplifies the algorithm design effort. Since linearization and noise assertions are not violated by the PF approach, this method potentially affords a more application robust solution strategy. Finally, a hardware implementation of the PF is presented as a practical mechanism of PF deployment that compensates for the otherwise computationally burdensome requirements imposed by PF random sample generation. This paper constitutes a preliminary effort conducted by SPAWAR Systems Center San Diego investigating the feasibility of PF application to GPS / INS navigation. Consequently, only a static receiver model is evaluated by application of simulation trials. However, even for this basic of navigation models, PF is demonstrated to provide an attractive alternative to the EKF.
Published in:	Proceedings of the 17th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2004) September 21 - 24, 2004 Long Beach Convention Center Long Beach, CA
Pages:	1619 - 1626
Cite this article:	Dyckman, Howard L., Sloat, Shon, Pettus, Bill, "Particle Filtering to Improve GPS/INS Integration," Proceedings of the 17th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2004), Long Beach, CA, September 2004, pp. 1619-1626.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In