A PDR Smartphone Application Considering Side/Backward Steps

Abstract

PDR is mentioned as a powerful indoor position estimation method. In general PDR, the direction of movement is determined from the value of the angular velocity sensor, so it is difficult to estimate behavior such as walking sideways or backwards without changing the angular velocity value. In order to solve this problem, we proposed a method to estimate the traveling direction of each step using the acceleration plane component. Although this research has made it possible to estimate the behavior in which the value of angular velocity does not change, there is a problem that the estimated trajectory becomes zigzag when going straight. The main reason for the zigzag trajectory is that the rotation of the waist is not taken into consideration. In this research, in addition to the acceleration plane component, the angular velocity plane component is used in combination, and the motion direction estimation for each step is made more accurate by canceling the waist rotation. As a experimental result, the error of the method of the previous research is about 14.2 degrees on average, and the method of this research is about 9.1 degrees on average, and the accuracy improvement of about 5.1 degrees on average was confirmed. In addition, the proposed method is implemented as an Android application with real-time processing enabled.