State Estimation Filtering Algorithms for Vehicle Attitude Determination using a Dual-Arc Accelerometer Array and 3-Axis Rate Gyroscopes

Abstract

Sensor measurements are corrupted by biases, noise and drift effects and, in order to provide accurate measurements, these errors need to be estimated and, thus, eliminated. The current model used an Extended Kalman filter for the estimation of rate gyroscope measurement errors. This work improves upon that filter by applying a more robust, more accurate and more reliable Unscented Kalman filter. In addition, an algorithm for estimating the accelerometer measurement errors is developed using control theory. Using the attitude estimate from the Unscented Kalman filter, an error signal is formed between that attitude and the attitude estimates from the accelerometer array(s). This error signal is then reduced by implementation of an innovative method using PID controllers to estimate, and reduce the effects of, accelerometer measurement errors. While this thesis uses a previously developed device and equations, it is a departure from the previous works as it considers parameters and variables that were ignored in those studies.