Design and Implementation of an AHRS Based on MEMS Sensors and Complementary Filtering

Abstract

This paper presents design and implementation of an attitude and heading reference system (AHRS) based on low-cost MEMS sensors and complementary filtering (CF). Different from traditional solutions, information fusion is performed with Euler angles directly, which is more straightforward for understanding; however it proposes many challenges for reaching a stable and accurate estimation as when these angles approach or traverse their range boundaries, estimation may get discontinuous. Thus an effective discontinuity avoiding strategy is suggested in this paper to refine the estimation. Besides, instead of extended Kalman filtering (EKF), CF is utilized for state estimation of AHRS as it features fusion of high-frequency and low-frequency signals. In order to make up for shortcomings of MEMS sensors such as multiple errors, drifts, and bad accuracy, some effective calibration and filtering algorithms are proposed to guarantee agreeable AHRS performance. Also, architecture of the MEMS IMU (inertial measurement unit) and mathematical principles for AHRS solution are explained and implemented in this paper. Meanwhile, experimental comparisons have proved feasibility and acceptable performance of this AHRS design.

Fast Linear Attitude Estimation and Angular Rate Generation

“…position, velocity and timing information [1], [2], [3]. The other is solving attitude solution by observation vector pairs matching, typically magnetometers, accelerometers and rate-gyroscopes (MARG) sensors [4], [5], [6]. In aerospace engineering and related geodetic applications, the latter is much more preferred avoiding the strong state parameters coupling existing in the former one [7].…”

“…In aerospace engineering and related geodetic applications, the latter is much more preferred avoiding the strong state parameters coupling existing in the former one [7]. For an attitude measuring system with MARG sensors, the sensor outputs usually contain the system noises due to internal and external factors [4], [5], [8], [9]. Therefore, an efficient fusion method is essential for integrating those complementary sensors.…”

“…Therefore, an efficient fusion method is essential for integrating those complementary sensors. A common approach to multi-sensor data fusion is complementary filters [5], [10], [11]. The frequency domain based filter structure allows for a straightforward implementation without requiring high performance signal processing units [12].…”

“…Combination of accelerometers and magnetometers is popular in compensating for gyroscope drift [8]. Alternatively, fusing accelerometers and magnetometers with gyroscope readings can greatly exempt the noise influences from attitude solution [5]. For the aspect of computation efficiency, many representative methods have been developed, like Gradient Descent Algorithm (GDA) [11], Gauss Newton Algorithm (GNA) [17], [18], Levenberg-Marquadt Algorithm (LMA) [31], [32], etc.…”

AHRS Design Based on STM32 and MEMS Sensors

“…while it is liable to be disturbed by other inertial forces, vibrations,or magnetic noises. Accelerometer and magnetic sensor could provide two uncorrelated estimation [7]. Gyro can follow the attitude angle change quickly and closely and cumulative errors can be corrected by using the observation datas of accelerometer and magnetometer.…”