| Summary: | 碩士 === 國立成功大學 === 測量及空間資訊學系 === 102 === The highly accurate mechanized platform inertial sensors without any external aiding signals had already achieved the positioning accuracy of 100 m in 1970s. However, this type of inertial sensors is very bulky and power consumption. In addition, these inertial measurement units (IMUs) are complex in the batch fabrication due to multitudinous parts so the cost has been always high. Despite the expensive IMUs, the accuracy requirements only conform to the applications. For instance, the IMUs applied in the civil aviation don’t need the excellent accuracy for free inertial since the radio receivers are integrated with the IMUs. The new materials, reduction of components, and innovative techniques for the inertial sensors are able to decrease the cost as well. Cold atomic interfered (CAI) IMUs based on their bias stability and random walk can provide significantly improved positional accuracy (〈10 m) than all current IMUs. Furthermore, the prices of CAI IMUs can be less expensive than current mechanized platform and optic inertial sensors on the demand of same accuracy. Nevertheless, whereas CAI IMUs are the underdevelopment techniques, they have some problems such as higher volume, lower sampling rate, and lower operation range, than traditional IMUs, obstructing practical applications.
In addition to the introduction of the operation principle, historical reviews, and related applications of CAI IMUs as well as the developed integration algorithms for the CAI IMU focusing on the problems of low operation range and low sampling rate, the developments and fundamental operation methods of famous conventional IMUs are also looked back in order to connect the relevance of inertial sensors in this research.
The adopted algorithms in this thesis include the generation of dynamic trajectories, measurement generation and the mechanization equations of inertial navigations, simulation of Global Positioning System (GPS), and the integration between the IMUs and GPS. The combination of dual IMUs (a micro electro mechanical system (MEMS) IMU and a CAI IMU) tries to utilize the high sampling rate of the MEMS IMU to assist the CAI IMU to capture detailed dynamics and restrain the positional drift caused by the MEMS IMU through the correction from the CAI IMU. The positional error of the MEMS IMU simulated in this research can achieve 300 m~400 m in one minute. The CAI IMU/GPS integration exploits GPS to reduce the positional error of the CAU IMU because of the inadequate sampling rate. Above two integrations only consider the problem of sampling rate. Nonetheless, the low operation range of CAI IMUs will produce serious positional error as the CAI IMUs are operated in high dynamic environment. The triple integration of a CAI IMU, a MEMS IMU, and GPS is capable of compensating the outages of the CAI IMU via the output of navigation solutions of the MEMS IMU aided with GPS.
According to aforementioned approaches, the integration between a CAI IMU and a MEMS IMU via loosely coupled scheme doesn’t have significant variation compared with free inertial CAI IMU since the states computed from the CAI IMU already containing sampling error will be fed back to the states of the MEMS IMU via updating. The GPS-aiding CAI IMU can reach good performances the horizontal and vertical accuracy of 5 m by the position, velocity, and position/velocity updating if no GPS outages. Apart from the case of no GPS outages, the estimated states of the CAI IMU/GPS is similar to the states of the unaided CAI IMU during the GPS outages supposing the measurement covariance matrix and system covariance matrix in the Kalman filter are proper for matching the conditions of the CAI IMU/GPS system. The programs of the triple system fulfill that the positional error aroused by the MEMS IMU can be controlled by the CAI IMU encountering the GPS outages and remained by the estimation of the MEMS IMU updated by GPS in CAI IMU outages. Whether the GPS outages take place or not, the positional accuracy of the triple system is a bit less than the CAI IMU/GPS system since the sensor error of the MEMS IMU will be propagated to the estimated states of the CAI IMU/GPS, too. Having said that, the two-stages updating in the Kalman filter can smooth the error fluctuation due to the incorrect estimation of the measurement noise variance matrix and the system noise variance matrix.
|
|---|
