Smartphone sensor based algorithms for Dead Reckoning using magnetic field sensor and accelerometer for localization purposes

Abstract

Dead Reckoning has successfully been used as a navigational method in the maritime field for positioning between two fixed locations. This approach can be also implemented for smartphones using acceleration sensor and magnetic field sensor. The reliability of the compass values obtained is important in order to calculate the new position. Different tests of accuracy of compass readings of two different smartphones under various test conditions are presented. Moreover, the accuracy of the reconstruction of two walked indoor/outdoor paths using Dead Reckoning will be investigated.

Smartphone-Based 3D Indoor Localization and Navigation

“…Similarly, the building's floorplan is to be considered as well, not only for visual representations, but also for determining valid movements, and for navigation indoors. Besides discussing all required theoretical mathematical backgrounds to determine implications and potential limitations of each individual component, the following scientific contributions will be provided throughout the course of this work: Probabilistic Sensor Models While dead reckoning [Ser28; ND97], pedestrian dead reckoning [Li+12;Cas+14], step-detection [Goy+11; TS12; SD16; PHP17; Kir+18], and activitydetection [Elh+14;Zho+15;Zha+18a] all are well-established fields of research, only few works focus on predictions that rely solely on a smartphone. Holding the device upfront, e.g.…”

“…When estimating this axis using just the smartphone's sensors, results can be unstable, and are subject to delays [Kus+15]. The heading provided by the eCompass is usually off by several degrees [Cas+14]. Using the phone's estimated orientation for projection thus is a viable choice for smartphone-only indoor localization and especially navigation, where the pedestrian faces the phone's display, similar to holding an analog compass.…”

Real-time PDR based on resource-constrained embedded platform

“…. , {" .. ; " & }a (9) ℎ \ = ! ℎ , ℎ $ , … , ℎ b - (10) ℎ * = " * E ; " * N (11) where " * E is the index at the beginning of a cluster and " * N is the index at the end of the cluster.…”

“…Distinctively, the third row denotes differences in time for both phases. The stationary phase took a longer time to complete because the number of samples to be processed in Equation (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13) equals to the window size, 40. This is a result of Equation (3) where all the acceleration samples at stationary phase are within the tolerance.…”

“…For example, Zheng et al [5] added a particle filter (PF) to enhance the position estimation of the pedestrian based on Kalman filter's output. Meanwhile, Zhang et al [6] attempted to replace double integration for distance conversion by using step length estimation, often used in smartphone's positioning techniques [7]- [9].…”