On Improving Navigation Accuracy of GPS/INS Systems

Direct georeferencing, also referred to as direct platform orientation (DPO), is defined as direct measurement of the imaging sensor external orientation parameters (EOP), using positioning and orientation sensors, such as the Global Positioning System (GPS) and Inertial Navigation System (INS) or Inertial Measurement Unit (IMU). Imaging sensors, most frequently supported by the DPO technique, are digital cameras, lidar systems, multi-spectral or hyper-spectral scanners, or interferometric synthetic aperture radar (INSAR). While for scanning sensors the use of DPO is compulsory, frame digital cameras can also directly benefit from this modern technique of sensor orientation. With direct sensor orientation, the requirement for ground control, tie-point matching and aerotriangulation (AT) is significantly reduced, or even entirely eliminated, resulting in shorter times of data acquisition and processing, and streamlined and highly automated data workflow and quality control. Most of the time, the requirement for ground control points is limited to periodic system calibrations and quality control check. Direct georeferencing is considered a fundamental technology of conventional mobile mapping systems (MMS). Despite significant progress in GSP/INS-based direct georeferencing technology in the last decade, there is still room for improvement in terms of better accuracy and better tolerance to GPS outages. This paper is focused on three error modeling and compensation techniques that could potentially improve GPS/INS system’s performance on both land-based and airborne platforms: (1) extended gravity compensation, (2) IMU signal de-noising, and (3) stochastic modeling of IMU errors. Introduction Integration of GPS and inertial navigation techniques is the primary means for direct platform orientation (DPO), providing high-accuracy timing, positioning and attitude information of the imaging sensor(s) mounted either on a land-based or airborne platform, thus forming a mobile mapping system (MMS). Real-time or post-processing of GPS/IMU data directly provide image georeferencing in a selected mapping coordinate system. Oriented images are then used in various photogrammetric processes to extract thematic and topographic information, such as terrain data, natural and man-made objects, etc. These features with their positional information and additional attributes can then be directly transported On Improving Navigation Accuracy of GPS/INS Systems Dorota Grejner-Brzezinska, Charles Toth, and Yudan Yi to a GIS database, or converted to a digital map. Also, with the increasing use of multi-sensor mapping or intelligencecollecting platforms, data fusion has become a crucial step in the design of these systems and an essential component of spatial data processing algorithms. The fundamental step of any data integration process is georeferencing or geometric fusion of data (time-space registration), provided by the Global Positioning System (GPS) and Inertial Navigation System (INS) or Inertial Measurement Unit (IMU): GPS/IMU or GPS/INS. Clearly, georeferencing by fusion of GPS and inertial technology is the back-bone of any modern mobile mapping system, and substantial research effort has been devoted to extensive algorithmic developments, performance analysis and practical implementations of GPS/IMU or GPS and dead-reckoning systems (Bossler et al., 1991; Schwarz et al., 1993; Bossler and Toth, 1995; El-Sheimy et al., 1995; Skaloud et al., 1996; Abdullah, 1997; Grejner-Brzezinska, 1997; Toth and Grejner-Brzezinska, 1998; Grejner-Brzezinska 1999; El-Sheimy and Schwarz, 1999; Mostafa et al., 2000; Mostafa and Hutton, 2001). Most of the publications to date in the mapping and navigation communities are focused on integrating differential GPS with high-end, navigation-grade IMUs. Since the market price of these systems is still rather high, and the performance of the consumer-grade IMU sensors still does not meet the high accuracy requirements of the majority of mapping projects, the challenge to examine the applicability of lower-end IMU sensors to direct georeferencing, under the assumption that special signal processing algorithms and extended error models are applied, is long overdue. Thus, the primary objective of this paper is to demonstrate several methods for additional improvements in the GPS/INS performance with a special emphasis on the enhancement of standalone inertial navigation with medium and low-end IMUs. In addition, we also examined the possibility to improve inertial navigation of the navigation-grade sensors, using an improved gravity compensation procedure (Grejner-Brzezinska and Wang, 1998; Grejner-Brzezinska et al., 2003 and 2004a-b) and careful modeling of the sensor errors. From the MMS perspective, our primary objective is to improve the long-term sensor performance and to achieve better accuracy. This is an ongoing research topic, thus, we do not present any final findings and recommendations yet. Instead, the theoretical foundation of our methods and preliminary test data analysis are provided for different IMU sensors combined with differential GPS in a land-based scenario. The test data, both static and kinematic, were collected in several survey sessions on the Ohio State University (OSU) PHOTOGRAMMETRIC ENGINEER ING & REMOTE SENS ING Ap r i l 2005 377 Dorota Grejner-Brzezinska and Yudan Yi are with The Ohio State University, Department of Civil and Environmental Engineering and Geodetic Science, 2070 Neil Avenue, Columbus, OH 43210. Charles Toth is with The Ohio State University, Center for Mapping, 1216 Kinnear Road, Columbus, OH 43212. Photogrammetric Engineering & Remote Sensing Vol. 71, No. 4, April 2005, pp. 377–389. 0099-1112/05/7104–0377/$3.00/0 © 2005 American Society for Photogrammetry and Remote Sensing MMS04-123.qxd 02/1/04 5:05 AM Page 377