Comparison of Euler Estimate using Extended Kalman Filter, Madgwick and Mahony on Quadcopter Flight Data

Abstract

A magnetic and inertial measurement unit (MIMU) provides raw, real-time acceleration, angular velocity, and a measure of earth's magnetic field. By itself, this data is subject to significant noise, bias, and drift (without constant re-calibration). A data fusion algorithm can be applied to significantly reduce these errors. In the past, many approaches have been adopted for filtering gyroscope data with inertial measurements, and the most commonly used techniques are Extended Kalman filtering and complementary filters. Thus, this paper compares three methods: two complementary filters known as Madgwick and Mahony, and the Extended Kalman Filter (EKF). Simulation experiments are conducted using quadcopter data and results show that Mahony provides better orientation estimation than both Madgwick and EKF when using optimum parameters.