Waist mounted Pedestrian Dead-Reckoning system

This paper proposes a waist mounted PDR(Pedestrian Dead-Reckoning) algorithm using a low cost MEMS IMU(Inertial Measurement Unit). The PDR algorithm is consist of three algorithms which are step detection, step length estimation and heading estimation. The step detection is to detect a gait of pedestrian in walking. The step length estimation is to estimate distance of walking. The heading estimation is to find direction of walking. The PDR scheme divides two methods depending on position of mounted IMU where foot or waist mainly. This paper uses waist mounted PDR for convenience of easy implementation. Peak detection and zero crossing method are used for detecting step using 3D accelerometer data. Step length estimation based on non-linear model is applied and HDR algorithm is used for estimating the heading. To verify the effectiveness of this system, real-time system is implemented and demonstrated. Experimental results show accuracy of below 3% position error.