A Movement-Classification Algorithm for Pedestrian using Foot-Mounted IMU
Minsu Lee, Changook Park, Chloe W. Shim · Seoul National University Open Repository (Seoul National University) · 2012
In this paper we present a PDR (Pedestrian Dead Reckoning) algorithm which uses characteristic features to identify different motions and eventually develop a classification algorithm for such motions. We used a single IMU sensor which imbedded at the pedestrians right foot. PDR is a device which provides position information of the pedestrian. Possible area of application for PDR is detecting the motions of rescuers or squads and locating their real-time spatial coordinates to enhance their rescuing and escaping. However, PDR in its original configuration and algorithm is not able to differentiate distinct motions such as running from plain walking, a function is required to accurately detect the motion of firefighters, who frequently crawl, climb up and down, squat, and run depending on the situation. Thus, we developed an algorithm for distinguish such motions using some significant features. Some tests are conducted in laboratory and verified result of our algorithm.