Sensors Data Fusion through Kalman Filtering

Abstract

Low cost MEMS inertial sensors are being widely manufactured and used in numerous applications ranging from defense, aerospace industry, agriculture, image stabilization, precision drilling, safety air bags in cars, mining, gaming, safety and many more. MEMS inertial sensors based low cost INS take the ad vantage of low price but performance achieved from such sensors is poor. GPS integration with such MEMS sensors improves their accuracy significantly. Such GPS aided MEMS inertial navigation systems are used in many safety applica tions such as road safety and monitoring. GPS and INS systems are usually integrated using a loosely coupled integration scheme, which is easy to imple ment. In case of GPS outage even for few tens of seconds the error grows very quickly. Such situation can arise in big cities with tall buildings, temporarily blocking GPS signals. Challenge, in such situations where GPS outage severely affects the performance, is to develop some sort of mechanism which minimizes quick error growth. This thesis investigates the performance of low cost MEMS based INS integrated with GPS in loosely coupled manner. LKF, EKF and UKF schemes are used in this work. MATLAB simulations are performed and scripts for LKF, EKF and UKF are developed. Further, a prediction method(PM) is proposed which improves the performance of INS/GPS during GPS Outage by minimizing the quick error growth.