Estimating Biases and Scale Factors in Speed and Yaw Rate Sensors Using a Linear Neuron

Faroog Ibrahim, Tom Pilutti, Nizar Al-Holou, Mark Paulik

Abstract:	This paper presents a new method for integrating INS and DGPS based on the use of a linear neuron (adaptive linear combiner element). The linear neurons in this work, learn adaptively, separately, and in a controlled way to estimate the errors in vehicle heading angle and speed. The neural network design presented in this paper is based on a mathematical modeling approach that decouples the heading angle error from the error in the forward speed. This allows errors to be estimated rather than approximating navigation models, and to easily extract the scale factor and the bias from the neural network connection weights. Using this approach, the neural network adaptively estimates the scale factor and the bias values during the availability of DGPS, and can then use these estimated values to aid the INS during DGPS outages or unsuitable DGPS solutions. The methods used to design and train the neural network, and the approach developed to estimate the desired targets are fully explained in this paper. Experimental results demonstrate outstanding performance for the proposed system compared to Kalman filter implementation.
Published in:	Proceedings of the 12th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1999) September 14 - 17, 1999 Nashville, TN
Pages:	2217 - 2226
Cite this article:	Ibrahim, Faroog, Pilutti, Tom, Al-Holou, Nizar, Paulik, Mark, "Estimating Biases and Scale Factors in Speed and Yaw Rate Sensors Using a Linear Neuron," Proceedings of the 12th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1999), Nashville, TN, September 1999, pp. 2217-2226.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In