Quaternion-Based Kalman Filter for AHRS Using an Adaptive-Step Gradient Descent Algorithm. (24th September 2015)

• Record Type: Journal Article
• Authors: Wang, Li; Zhang, Zheng; Sun, Ping
• Is Part Of: International journal of advanced robotic systems. Volume 12:Number 9(2015)
• Publication Date: 2015-09-24
• Subjects: AHRS -- Kalman Filter -- Gradient Descent -- Inertial Sensors -- Magnetic Sensors -- Orientation Estimation -- Quadrotor -- Quaternions; Robotics -- Periodicals; Robotics; Periodicals; 629.892
• DOI: 10.5772/61313
• Languages: English
• ISSNs: 1729-8806

Abstract

This paper presents a quaternion-based Kalman filter for real-time estimation of the orientation of a quadrotor. Quaternions are used to represent rotation relationship between navigation frame and body frame. Processing of a 3-axis accelerometer using Adaptive-Step Gradient Descent (ASGD) produces a computed quaternion input to the Kalman filter. The step-size in GD is set in direct proportion to the physical orientation rate. Kalman filter combines 3-axis gyroscope and computed quaternion to determine pitch and roll angles. This combination overcomes linearization error of the measurement equations and reduces the calculation cost. 3-axis magnetometer is separated from ASGD to independently calculate yaw angle for Attitude Heading Reference System (AHRS). This AHRS algorithm is able to remove the magnetic distortion impact. Experiments are carried out in the small-size flight controller and the real world flying test shows the proposed AHRS algorithm is adequate for the real-time estimation of the orientation of a quadrotor.