Adaptive Kalman Filter for IMU and Optical Incremental Sensor Fusion

Abstract

The main problem of the use of inertial measurement unit (IMU) lies in the systematic increase of position and orientation errors resulting from their functional principle. Position and orientation calculation based on the integration of inertial measurements (accelerations and angular velocities) leads to a rapid accumulation of errors of the inertial sensor with increasing time of measurement. Existing solutions for the elimination of these errors focus on a combination of IMU with several additional sensors to increase long-term positioning and orientation accuracy. The article deals with the elimination of systematic errors of IMU, based on optical incremental sensor measurements and conditions resulting from the actual kinematic state of IMU. Incremental encoder measurements are used for the correction of speed and travelled distance. The proposed processing model uses the automatic identification of the kinematic state of the IMU, which is realized based on inertial measurements. Based on the current kinematic state of IMU, the process model has been modified, which allows a better way of modelling and eliminating systematic error. The verification of the efficiency of the designed model is realized by experimental measurements. During experimental measurements, the measuring system (IMU and optical encoder) is installed on the trolley. The predefined trajectory of the trolley’s movement is given by the known position of reference points, based on which the differences in distance, the orientation differences, the positional differences and their development with increasing time of measurement are analyzed.