보행자의 불특정 동작이 미치는 PDR 성능 분석에 대한 사례 연구

Abstract

This paper describes the case study on the performance analysis of PDR (Pedestrian Dead Reckoning) by unspecific motion of pedestrians. PDR is a device which provides position information of the pedestrian. In general, the step length is estimated using a linear combination of the walking frequency and the acceleration variance for mobile devices. Step length estimation performance is evaluated when the pedestrian looking around while walking. A situation is assumed that the subject is looking around while walking and take wavering steps in a specific spot. In this situation, unspecific motions which are walk or walk in place or walk back and forth in a short distance or stop for a moment can be occurred. The step length estimation algorithm with a walking status awareness algorithm is tested in this situation. This test was verified by two different subjects and the estimation error of this case was up to 7.6% at the worst case.