Research On The Indoor 3D Positioning System Based On Mems Inertial Sensor

Driven by the development of MEMS technology, as well as the urgent demand for location based service, the indoor inertial positioning system has become a hot spot at present. This system with its advantage of independent navigation, no additional infrastructure or network facilities, and the ability of working under complex environment is very worth to do a deeper research. Nowadays, most indoor inertial positioning systems adopt commercial MEMS inertial sensor, which is expensive, and only can be used for 2D space positioning, not suitable for popularization and application. Hence, aiming at all these disadvantages, this paper proposes a 3D indoor positioning system based on a type of MEMS inertial sensor with low cost, which can realize 3D space positioning and expand the range of the indoor positioning system.In this paper, the hardware circuit of 3D indoor positioning system is made up of three parts. The MEMS inertial sensor MPU-9150 and the pressure sensor BMP-085 are used as an original data acquisition unit. The other two parts are a control unit dominated by the STM32 and a wireless Bluetooth serial port to transmit data. Firstly, initial Euler Angle is obtained by using the accelerometer and magnetometer. After that, with relation of Euler angle and quaternion, the initial quaternion is obtained. Secondly, a parallel multi-sensor information fusion algorithm is used to calculate heading and attitude angles. Further-more, through the cooperation of the accelerometer and gyroscope, the stability and the accuracy of this algorithm can be improved. Thirdly, the accuracy of attitude detection can be up to 98% with a double threshold algorithm instead of a single threshold. Meanwhile, the ZUPT self-adapting step length calculation method is utilized to calculate the real walking step length and trajectory. With its advantage, the complexity of the step length calculation and the integral drift error can be reduced. Finally, this paper proposes an effective method to eliminate the heading drift error.With actual tests and analysis, the results are as follows: the accuracy of attitude detection can reach to 98%. The error of the vertical height is about 1m. The positioning error is about 10.1% in the total travelled distance when walking in a line, deviating from the straight direction is 8°, the error is about 12.56% when walking back and forth in a straight line, and the heading error is about 50°, the error is about 11.67% when walking in a rectangle. Based on these, a low cost inertial 3D indoor positioning system can be achieved.