Improved Position Estimation of Real Time Integrated Low-Cost Navigation System Using Unscented Kalman Filter

Abstract

Low cost Micro Electrical Mechanical Systems (MEMS) inertial sensors have nominated their use in domains such as navigation systems but these sensors are noisy and characterized by their measurements drift and large errors. In this work, an integrated navigation system is implemented and its performance is evaluated through experimental work in both post processing and real time domains. The real time processing is built on multi-platform 32-bit ARM core ATMEL microcontroller while at the same time raw sensors measurements are saved for post processing under MATLAB. North and East position errors measurements were used in this work to evaluate position improvement calculated using MEMS inertial sensors integrated with position and velocity from Global Positioning System (GPS) receiver readings at slow rates. For such evaluation, the calibrated measurements from gyroscopes and accelerometers are fed to a mechanization process to build an inertial navigation system (INS). The INS drifting navigation solution is fused with a GPS measurement through Unscented Kalman Filter (UKF). UKF does not require linearization of the system model such as the status of the commonly used Extended Kalman Filter (EKF). The results of the experimental work show that the implemented low cost integrated navigation system based on UKF can achieve a level of accuracy superior than other (EKF) based expensive systems.