This Is AuburnElectronic Theses and Dissertations

Performance Analysis of Attitude Determination Algorithms for Low Cost Attitude Heading Reference Systems




Narayanan, Karthik

Type of Degree



Electrical Engineering


Development of micro-electro mechanical system (MEMS) and micro-electro optical mechanical system (MEOMS) inertial sensors has been driven by the need for inexpensive sensing solutions in military and commercial applications. In addition to traditional attitude estimation and automobile applications, the reduced cost of MEMS/MEOMS inertial sensors has spurred new applications in personal navigation, pose estimation, audio visualization, cueing, etc. Electromechanical inertial sensors are also increasingly used in low cost attitude heading reference systems (AHRS) and backup attitude indicators, when the required accuracies are not too stringent. With the current performance levels of MEMS sensors, AHRS systems using electromechanical sensors rely on some form of external aiding to generate a better attitude solution. External aiding could come from an air data computer, global positioning system (GPS), etc., but the aiding comes at an increased cost. Aiding sources have their own set of errors and may not be available at all times. For example, air data sources su er from problems such as icing, blocked pressure ports etc., and GPS integrity can be compromised due to interference. This dissertation addresses the problem of low cost attitude estimation using a triad of MEMS gyros and accelerometers for xed wing and rotary wing aircrafts, under conditions when external aiding is unavailable or not useful. Using angular rate, acceleration and magnetic measures, a unit quaternion algorithm is formulated by combining a non linear attitude estimator with fuzzy logic concepts. Static and dynamic conditions of the aircraft are exploited to adaptively alter gains in correction loops used to correct input rate measures. Standard tests are simulated to assess the performance of the formulated algorithm. Real world ight data is used to compare the results of the proposed algorithm with an extended Kalman lter, and the error analysis is presented. It is shown that the fuzzy non linear estimator algorithm can be used to compute a reasonably accurate attitude solution using inexpensive MEMS sensors, even when external sources of aiding are unavailable.