Development of High-precision Navigation Algorithm by Fusion of GPS and MEMS Sensors

摘要

This paper presents development of highly accurate navigation algorithm by sensor fusion, integrating information of civil-use GPS receiver and several MEMS sensors. It intends to make up navigation equipment generally used for small moving objects such as robots and Unmanned Aerial Vehicle (UAV). Then it is necessary to use small, light but low-accuracy MEMS sensors as measurement devices for achieving the goal, and therefore this research provides algorithm to integrate them effectively and constructs more accurate, highly-robust and reliable navigation system than previous researches by utilizing many kinds of MEMS sensors and the INS/GPS integrated navigation system. A civil-use GPS receiver, two different-range tri-axis gyroscopes and accelerometers, a tri-axis magnetic sensor and a pressure sensor are mounted on the navigation equipment, and information obtained by all of them is integrated by extended Kalman filter. As for combining different-range sensors, it is possible not only to enlarge measurement range simply, but also to improve quality of the signals by focusing on their frequency responses and filtering methods. This paper introduces the innovative way to treat these multiple sensors. Also, refined methods of integrating a magnetic sensor and a pressure sensor into the navigation system are described. In the latter, functionality of the algorithm is validated with computational simulations with aircraft navigation data actually obtained in flight experiments. Results of the simulations show that the magnetic sensor enhances accuracy and reliability of attitude estimation especially in the situation that the motion of an aircraft is so little that attitude estimation error can be accumulated without a magnetic sensor. They also show that the pressure sensor is effective for the overall system to acquire robustness of estimating altitude and vertical velocity especially in the situation that GPS signals are not available.