SpinMag: A New Fingerprinting Method for Robot Indoor Localization with Geomagnetic Field
Ruochun Jin, Kui Wu, Yong Dou and Mantis Cheng
Indoor localization, especially indoor localization for mobile robots, has become a crucial technology in the last two decades because of the proliferation of robots in industry and people’s daily life. We develop a novel approach to robot indoor localization, using fingerprint features extracted from the geomagnetic field by spinning the compass sensor in a smartphone. We evaluate the system with respect to the error distance and the accuracy rate, with the former capturing the distance between the real position of a robot and the reported position by the system, and the latter capturing the probability that the system successfully locates the robot within a given error distance. Comprehensive experimental test shows that our method can achieve the average error distance of 1.2 meters and 68% accuracy rate with the tolerable error distance set to 1.2 meters. This result outperforms a state-of-the-art solution introduced by Subbu et al in 2013.
Keywords: Indoor localization, robot navigation