Previous Abstract Return to Session D2 Next Abstract

Session D2: Robotic and Indoor Navigation

Indoor Localization for Pedestrians with High Accuracy and Usability Based on Smartphones
Sheng Yang, Jingbin Liu, Zhenbing Zhang, Xiaodong Gong, Gege Huang and Yu Bai, State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, China
Location: Galleria I/II
Alternate Number 2

Location-based services and navigation have been studied over the past few decades, which can bring great convenience to people’s daily lives, such as personal navigation, facility management, object search, disaster prevention and robot control. Smartphone, as popular cost-effective multi-sensor device, is a promising platform to provide reliable location information. Due to the complexity of the indoor environment, it is still a challenge to realize reliable and high accuracy positioning solution. In order to address this problem, many different techniques have been explored by researches, such as wireless local area network (WLAN), Bluetooth, magnetic field matching, ultra-wideband (UWB), visual positioning and pedestrian dead reckoning (PDR). These methods are often adopted under certain conditions due to their own characteristics. For example, fingerprint-based techniques are suitable for the corridor environment and will become powerless in the large rooms. Bluetooth antenna array systems can achieve centimeter-level accuracy which need additional infrastructures resulting in higher cost. The visual positioning method requires sufficient ambient light and image features to get better results, and the positioning error of PDR accumulates along with time. Therefore, many sources with complementary characteristics are often combined to improve the accuracy and stability of the positioning system.
In this paper, a hybrid indoor localization system based on smartphone sensors is proposed which can achieve full coverage, high accuracy and real-time positioning. It utilizes high-precision sources, i.e., sound source and light source, to maintain decimeter-level precision in the hall and the rooms, employs smartphone’s built-in inertial sensors as a link between different rooms, and intelligently fuses WLAN and magnetic field fingerprinting to minimize the accumulation error. Specifically, centimeter-lever accuracy can be achieved in the hall under the control of the light source, and decimeter-lever accuracy will be reliably attained in the sound source’s rooms. The PDR method is adopted as a link between different rooms, and its accumulation error is suppressed under the correction of WLAN and magnetic source. The novelty of this research refers to greatly extend the PDR positioning performance and correct its heading angle and speed drift once entering into the high-precision sources’ areas. This system also enables expansion for different applications, so other existing position methods can be easily added into it. The multi-source fusion scheme is based on the extended Kalman filter (EKF) and Unscented Kalman filter (UKF); the former estimates the angle and angle rate using acceleration, gyroscope and magnetometer data, and the latter plays the primary filtering to get an excellent performance. In the filtering process, the statistical means are used to carry out multi-source mutual inspection and gross error elimination with the aim of ensuring the stability and reliability of the hybrid positioning solution.
To verify the performance of the proposed method, extensive field tests were conducted at the office building which has an area of about 1500 square meters and is located in Wuhan, China. The presented method resulted in accuracy of 1 meter (1 sigma) in localization distances of the entire floor, and showed the improved performance in terms of accuracy and usability compared to the separate positioning method.



Previous Abstract Return to Session D2 Next Abstract