An Initial Alignment Method of Mobile Robot Rotational Strapdown Inertial Navigation System Based on RWNN Observer

Abstract

The positioning performance of intelligent mobile robot is directly affected by the performance of inertial navigation system (INS). In order to improve the long-term performance of low precision MEMS level navigation system, an initial alignment method of mobile robot rotating strapdown inertial navigation system based on RWNN observer is studied. In this method, the non random errors in gyroscope and accelerometer are eliminated through the integration operation of strapdown inertial navigation algorithm in rotational base, and the platform error angle is quickly estimated based on the filtering method of the RWNN (Recurrent Wavelet Neural Network) observer, so as to complete the initial alignment of INS. This paper discusses the error dynamics state space model of RSINS, as well as the construction process of the RWNN observer, and verifies the performance of INS based on a certain type of fiber-optic gyro and MEMS accelerometer. The result of experiment result shows that the initial alignment method proposed in this paper has high alignment performance, and can be widely used in new energy power inspection robot, automatic driving positioning and other relative fields.