An Improved Dead Reckoning Algorithm for Indoor Positioning Based on Inertial Sensors

Location based service (LBS) became more and more popular. The core technology of LBS is positioning. For indoor positioning, the GPS (Global Positioning System) receiver can't provide accurate results and even fails in positioning. This paper raises a method by using inertial sensors to realize a continuous indoor positioning. The method used Pedestrian Dead Reckoning (PDR) algorithm, which utilized accelerometers and gyros to determine step, stride and heading. In common PDR algorithm, the step is counted when the measured vertical impact is larger than a given threshold value. Sometimes steps are miscounted because the vertical impacts are not correctly calculated due to inclination of the foot. This paper proposes a new method for step detection and stride estimate: The stride is not constant and changes with speed, so the step length parameter must be determined continuously during the walk in order to get the accurate travelled distance. The new step detection method uses pattern recognition which proposed from the analysis of the vertical and horizontal acceleration of the foot while walking. A stride estimate method is also obtained by analyzing the relationship between stride, step period and acceleration. Heading is output from gyros and magnetic compass. Experiments showed that the result of new method is more effective and reliable.