| Abstrakt: |
Despite the existence of various sensor errors and users' complex movements, smartphone sensor-based positioning systems have significant potential for indoor location-based services (LBS). In this paper, a method for indoor positioning in three smartphone carrying modes (i.e., Texting, Calling, and Swinging) using accelerometer, gravity, magnetometer, gyroscope, and pressure sensors data is proposed. The data of the gyroscope, gravity, and magnetometer sensors are integrated to estimate the heading using the Kalman filter (KF) based on the quaternion of the orientation of the phone. To evaluate the heading estimation method, five different methods including the magnetic (combined with the accelerometer and gravity sensors data), Madgwick, Extended Kalman filter (EKF), and proposed KF are implemented independently for different carrying modes. Gravity and magnetometer information is used to update the gyroscope state model. Finally, the 2D position is estimated based on the length and heading information of the steps. To develop the system for 3D positioning applications, the pressure sensor of the smartphone is employed to measure the changes in the pressure related to the vertical displacement of the user and the changes at the floor level. However, this method as a relative positioning method suffers from the cumulative error problem, which prevents its independent implementation. The heading and step length errors cause deviation from the main path of movement and cut off the walls. A new map matching method is introduced by examining the user's position and the way the path crosses through the wall to prevent the wall-crossing problem. The results show that the proposed algorithm significantly improves the accuracy of position estimation. The mean absolute reference position error for the three smartphone carrying modes (i.e., texting, calling, and swinging) was reduced from 3.43 to 0.59, from 3.56 to 0.67, and from 4.83 to 0.77, respectively. [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink