A New Approach of Real Time Step Length Estimation for Waist Mounted PDR System

Abstract

Pedestrian dead reckoning (PDR) is a promising solution for indoor positioning. It can work without either GPS signals or Wi-Fi coverage. Commonly, a foot mounted PDR system is relatively more accurate, but a PDR system mounted on the waist or integrated in a Smartphone can be more convenient. However, such systems have difficulties with step length estimation. Most of them estimate step length indirectly according to acceleration features. This paper proposes a new approach of step length estimation for waist mounted PDR system which does not require detection of zero velocity point. Our approach can provide 96.9% accuracy in walking distance estimation. In addition, we also discussed the influence of mounting position of sensors to the accuracy. Finally, we applied the principle of step length estimation to state detection, distinguishing normal walking, descending or ascending. Tests show that as high as 81%, our method can correctly distinguish the person's walking status with 81% accuracy.