Fuzzy based complementary filter design for attitude estimation

Navigation systems that provide attitude, position and speed information are widely used in different fields today. Ring laser and fiber optic gyroscope sensors for angular velocity measurement, Servo and Quartz accelerometer sensors for acceleration measurement are known as the most sensitive sensors. Although eligible for many applications, they are not suitable for use in small size unmanned air vehicle (UAV) systems because they are large and heavy. Moreover, the prices of these sensors are relatively high. In order to attitude estimate, micro electromechanical systems (MEMS) are widely used in UAV systems due to the fact that they are cheap, light-weight and small size. Though the advantages of MEMS based inertial sensors, they have some weaknesses that affect performance, such as low resolution, noisy output and poor bias stability. For these reasons, MEMS sensors such as accelerometers, gyroscope and magnetometers cannot provide sufficient navigation solutions as a stand-alone system. Therefore, the data from each sensor must be fused together. This thesis proposes a novel fuzzy based complementary filter providing the fusion of acceleration, angular velocity and magnetic field measurements obtained from MEMS based inertial sensor to provide more accurate attitude estimation under the dynamic motion. The complementary filter was preferred because of its distinctive features such as the use of a high-pass filter to reduce measurement noise, less computational complexity, and consideration of possible drift errors. However, the constant-gain complementary filter shows low sensitivity under dynamic motion. To overcome this problem, a fuzzy based algorithm has been developed that can make the complementary filter gain dynamic.