Vibration effects rectification of IMU attitude based on gradient descent algorithm

Recently the shift to low cost sensors has led to an exponential growth in the range of applications for inertial measurement. Almost all of these applications are situations where vibration is present at some level. Based on gradient descent, this paper presents a novel algorithm for inertial measurement units (IMU) attitude calculating of rectification in vibration environments. With inertial MEMS sensor which includes three-axis accelerometer, three-axis gyroscope, the euler angle can be computed by the method of quaternion which can reduce computing complexity and avoid Gimbal Lock in rotation. The data of accelerometer was used to derive and optimize gradient descent algorithm for calculating the quaternion derivative which representing the direction of gyroscope measurement error. The data was acquired from vibrostand equipped with IMU sensor in different models of vibration experiments, including single and dual axis fixed frequency sinusoidal vibration, sweep frequency vibration and random vibration. The accuracy of roll and pitch is average <; 0.5° rectification by the algorithm this paper presented.