Smartphone Based Indoor Navigation System Using Motion Recognition and Map Matching

B. Shin, C. Kim, J. Kim, S. Lee, C. Kee, T. Lee

Abstract: A motion recognition-based indoor navigation system that employs a smartphone is presented. The smartphone has various onboard sensors, such as an accelerometer, gyroscope, and magnetometer that can be used to operate the PDR system. As inexpensive and small sensors with low power consumption have been manufactured through the development MEMS, the IMU-based PDR system has become more accurate and easier to implement. The IMU-based PDR system is effective when used for specific purposes, such as tracking firefighters or soldiers. However, this system is inappropriate for pedestrian indoor navigation. A smartphone-based indoor navigation system is much more convenient than the IMU-based system for pedestrians. The pedestrian using the smartphone can access the indoor navigation service simply by downloading the application. In this research, the smartphone based indoor navigation system is presented using motion recognition and new map matching technology. In the case of using a smartphone, the pedestrian typically holds the smartphone in his or her hand and performs various motion such as standing, walking looking at the smartphone, walking talking on the device, and walking with the device in pocket. To estimate the motion of pedestrian, we make the motion recognition solution using DT and SVM. The DT is used to consider whether the pedestrian is static or moving through the variance accelerometer output. Also we utilize the proximity sensor because the proximity sensor can detect the presence of nearby objects. When pedestrian performs the talking and pocket motion, the proximity sensor is enacted. Also talking and pocket motion could be recognized by using variance value of accelerometer output. PDR is consisted of step detection, step length estimation and heading estimation. Accelerometer is used to detect pedestrian step occurrence by using peak detection method. The step length is estimated by using linear combination method. The heading of the pedestrian is the most significant factor in the PDR. The magnetometer and gyroscope are used in the heading estimation. Furthermore, the Kalman filter (KF) is a useful tool when estimating the heading. The magnetometer provides the absolute heading. However, it is easily disturbed by the surrounding environment. On the other hand, the gyroscope is not influenced by the external environment, while it is more sensitive than the magnetometer. In addition, its output is reliable in a short amount of time. However, it gives only the relative heading, and the error is accumulated over time. Therefore, the gyroscope is used for time update in KF, and the magnetometer is used for measurement update. When pedestrian walks with holding the smartphone, the axes of the smartphone move continuously. The direction cosine matrix is applied to convert the body frame vector to navigation frame vector. When the pedestrian changes the motion, the time update of KF is not carried out. The timing of motion transition could be detected through the motion recognition algorithm. Angular velocity generated by the motion change should not be integrated into the heading of pedestrian. Thus, before and after the motion transition, the gyroscope integration is skipped. And during calling or pocket motion, the axes of smartphone is different with the navigation frame. Thus, the bearing degree is added to heading estimated by magnetometer. Map information is useful to increase the positioning accuracy of navigation. We make the indoor map which is consisted of links and nodes. The link could be used to update the estimated heading of pedestrian. If it is reliable that pedestrian is on certain link, the link direction is used to the measurement update of KF. In case of using link direction than magnetometer, the covariance of measurement is reduced to estimate the heading more efficiently. If wrong movement of pedestrian is detected continuously, the position of pedestrian is moved to candidate node. And the map information is used to determine the resolution of heading direction. For example, the heading resolution is more enhanced in the corner than in the area like corridor. To verify the proposed system, we developed the android based indoor navigation application. This application has three main classes such as motion-awareness PDR and MM. The experimental results shows that the positioning error in the indoor environment is less than 5 meter during experiments. It is expected that the proposed system will be applied to various applications for smartphone-based indoor navigation.
Published in: Proceedings of the 28th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2015)
September 14 - 18, 2015
Tampa Convention Center
Tampa, Florida
Pages: 24 - 54
Cite this article: Shin, B., Kim, C., Kim, J., Lee, S., Kee, C., Lee, T., "Smartphone Based Indoor Navigation System Using Motion Recognition and Map Matching," Proceedings of the 28th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2015), Tampa, Florida, September 2015, pp. 24-54.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In