A Method for Estimating Walking Speed by Using Magnetic Signature to Grasp People Flow in Indoor Passages

The goal of our study is to grasp people flow velocity in indoor passages, by collecting the data of people's walking speeds and the area through which they have walked (walked section). In this paper, we propose a method that estimates the walking speed and walked section of a pedestrian by using a smartphone equipped with a magnetometer. To grasp people flow velocity, the walking speed of a pedestrian is an important scale. Besides this, to grasp low flow velocity spots such as an intersection and an area in front of a door, we need to estimate the walked section, that is, the area between two points that a pedestrian walked within a period of time. In our proposed method, first we measure magnetic field data series while a participant is walking, and we generate magnetic signatures by relating each magnetic field datum to the corresponding position. Secondly, we detect partial magnetic signature that is similar to the magnetic field data series of an unknown walked section by using DTW (Dynamic Time Warping), and we obtain the walked section from the detected partial magnetic signature. Finally, we calculate walking speed from the distance of walked section and walked time. We conducted an experiment to evaluate the accuracy of walking speed estimation and walked section estimation. The results in a crowded passage showed that the average error of walking speed estimation was 0.09 m/s and the average error of walked section estimation was 0.7 m in the proposed method.