A Low-cost, Loosely-coupled GPS/SINS Integrated Navigation System: Design And Implement

Recently, quadrotors have been extensively applied in various areas, such as military applications, traffic surveillance and logistics. Low-cost integrated navigation technology has become a spotlight in the commercial and academic areas, due to its key role on the autonomous flight of quadrotor. In practice, loosely-coupled GPS/SINS integrated navigation is easy to implement. However, much challenge limites the low-cost applications. Serious system errors and random noises of inertial sensors decrease the performance of navigation system. Multi-path effect in GPS outputs may lead the navigation filter divergent. In addition, statistical noises are not easy to estimate.This thesis proposes a low-cost loosely-coupled GPS/SINS integrated navigation system, emphasis on design and implementation. Firstly, the overall design and hardware scheme is described in accord with system metrics. After collecting data from the proposed platform and analyze in detail, the thesis gives the error model of inertial sensors. A simple fast field calibration algorithm is proposed to reduce system errors and a prefilter is designed to degrade the intensity of random noises. Based on the low dynamic characteristics of quadrotor, the thesis proposes a constant velocity model for GPS outputs which can be easily implemented easily using Kalman filter for prefiltering. Results show that the calibration methods and prefilterings both have a role to improve the performance of sensors and GPS.Further more, the error transfer equations of low-cost SINS are derived in the thesis, including perturbation equations of position, velocity and attitude. These equations are integrated to build state equation of the propsed integrated navigation system. In addition, in order to estimate the stastical feature of measurement noise precisely, an empirical model is presented to signify the relationship between the variance of magnetometer/barometer and motor speed. Flight test in the end show that, the proposed low-cost GPS/SINS integrated navigation system can improve the navigation performance, compared with single low-cost GPS or SINS.Different from the existing research, the main contribution of the thesis is summarized as follows:1. This thesis proposes a fast field calibration algorithm for low-cost inertial sensors. According to the real test, time-various biases of inertial sensors are no apparent, thus only fixed biases can be taken into account. The proposed calibration algorithm can correct biases and scale factor errors without and external equipment;2. Efficient prefilter is designed aimed at decreasing the impact of white noise, which is quantitatively analized using Allan variance. Based upon the low dynamic characteristic of quadrotor, a constant velocity model is used for Kalman failtering of GPS ouputs;3. To counter the low-cost navigation applications of quadrotor, a novel integration algorithm is proposed. The peculiarity dominately reflects in the relationship between motor speed and measurement noise.