Purely Inertial Navigation with a Low-Cost MEMS Sensor Array

Abstract

This paper examines the position precision of purely inertial navigation using an array of redundant, low-cost MEMS sensors. A carefully designed IMU is used to perform navigation experiments and to analyze the benefits of a sensor array over a single sensor in practice. As our experimental results show, navigation can be improved significantly by calibrating the IMU device regarding scale factors, offsets and cross-axis sensitivity. By comparing predicted navigation error and experimental results it is shown that gyroscope angle random walk and bias instability are dominant and therefore can be used to estimate naviaation performance. The latter improves roughly by a factor of $\sqrt{14}$ when using an array of 14 devices instead of a single one. A Kalman Filter with motion constraints minimizes the error when estimating positions.