GPS/INS系統整合與實驗

Abstract

在導航系統中，全球定位系統之應用漸趨廣泛普及。然而GPS於城市使用常受限於衛星訊號品質，當環境高樓遮蔽，造成衛星訊號不佳，使得最低需求無法滿足，可能導致GPS接收機無法提供即時準確之導航資訊，為解決此問題，我們藉由慣性感測器的自主導航特性，發展GPS/INS整合系統，彌補GPS訊號中斷之空窗期。 由於慣性感測器存在一偏置值，使得其定位誤差愈來愈大，我們利用延伸型卡爾曼濾波器於GPS訊號未中斷時估測慣性感測器的偏置值，在GPS訊號中斷時則使用校正後的INS進行定位，使整合系統可以克服個別系統單獨操作的缺點並提供更佳的導航資訊。 本文利用載具在平面上運動的方程式建立二維GPS/INS整合式導航系統，藉由模擬設計動態實驗，證實整合系統確能掌握偏置值及提高定位的精度。為檢驗我們所發展系統之適用性，本文整合GPS接收機、慣性感測器及電子羅盤，進行單接收機路測實驗，且依據路測環境曲率的變化，發展出可自我調變狀態雜訊協變異矩陣的適應性卡爾曼濾波器，提高系統定位的效能。經由實驗數據驗證，即使在短時間無GPS訊號的情況下，我們所發展的整合導航系統確實可達成定位推估之有效性及可靠性的目標。

The Global Positioning System (GPS) has become widely used in navigation systems. However, using GPS in the city is often limited by satellite signal quality. GPS receiver cannot provide precise and real-time positioning information when its signal is blocked. To deal with this problem, another independent navigation system INS (Inertial Navigation System) can be adopted and the GPS/INS integrated system may be developed to provide the positioning information during the GPS signal outage. Although INS is autonomous, the errors of using INS will increase with respect to time due to the drift problem in the integration procedure. By using the extend Kalman filter, the drifts of INS may be corrected by GPS signal. Alternatively when the GPS is outage, the INS can still provide positioning information. The integration system can not only solve the sensor’s problem but also enhance the performance of the navigation system significantly. In this thesis, the 2-D GPS/INS integrated system is developed for vehicles moving on the plane. The GPS/INS integrated navigation system design is first evaluated by dynamic simulation to confirm that the integration system can really grasp the bias error and improve positioning accuracy. In order to test the applicability, we proceed to the road test by integrating a GPS receiver, inertial sensors and an electronic compass. Moreover, the state noise covariance matrix in the Kalman filter is changed according to the curvature of the road to improve the performance. Such adaptive scheme enhance the performance of the algorithm is verified by the experimental results.