Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #1

DEVELOPMENT OF GPS AIDED INERTIAL NAVIGATION SYSTEM FOR HIGH SPEED FLIGHT DEMONSTRATOR:
M. Harigae, National Aerospace Laboratory, Japan; T. Nishizawa, National Space Development Agency of Japan; H. Tomita, TOSHIBA Corporation, Japan

NAL and NASDA develop High Speed Flight Demonstrator (HSFD) to evaluate the automatic take-off and landing technology and the transonic aerodynamic characteristics of the reusable space plane. A navigation system of the HSFD must meet the high-level requirements in its accuracy, integrity and continuity. The conventional DGPS and Kinematic GPS (CDGPS) navigation systems are not enough to satisfy the requirements of the precision approach and landing on account of the high dynamic motion of the HSFD.

This paper describes the development of a GPS aided inertial navigation system for the HSFD whose distinctive feature is using the carrier-phase DGPS (CDGPS)/INS hybrid navigation technology. The CDGPS/INS navigation is a novel approach to realize high navigation accuracy without lack of good integrity and continuity performance. Because the ambiguity of the carrier phase is resolved with the aid of INS information, the CDGPS/INS navigation can avoid the degradation of navigation performance when the GPS signal lock-loss and re-acquisition occur under the high dynamic motions.

Performance analyses are made on the accuracy and the integrity of the CDGPS/INS navigation for the HSFD missions. The covariance analyses assure that the accuracy of the CDGPS/INS navigation is 0.4 m (3s) in horizontal and 1.8 m (3s) in vertical under the worst GDOP case at Christmas Island where HSFD experiments are carried out. The integrity-monitoring algorithm of the CDGPS/INS navigation checks the measurement residuals (innovation process) to detect malfunctions in GPS data. Because of high navigation accuracy, the protection level (PL) of the CDGPS/INS navigation is very good. The risk analyses indicate the availability of the CDGPS/INS navigation is more than 0.999 even though one GPS satellite failure is considered.

To certificate the performance analyses, we carried out flight experiments. More than 30 times touch and go flights showed the positioning accuracy of the CDGPS/INS research model was 0.4 m (95%) both in horizontal and vertical in case of nominal GDOP conditions. The flight model of the CDGPS/INS navigation system is now under ground and flight tests. Its weight is about 10 kg and the size is 180 mm x 180 mm x 280 mm. The power consumption is about 60 W. The operating temperature is -40 degree to +55 degree (+71 degree for 30 min.). The operating altitude is 0~32 km.

The developed CDGPS/INS navigation system meets the requirements of navigation performance under severe conditions. The system shows highest-level performance among the developed GPS/INS systems. Its technology may be useful for manned general aviation that requires high reliability and tolerance to unintentional interferences.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #2

FDE USING MULTIPLE INTEGRATED GPS/INERTIAL KALMAN FILTERS IN THE PRESENCE OF TEMPORALLY AND SPATIALLY CORRELATED IONOSPHERIC ERRORS:
K. Vanderwerf, Honeywell, Commercial Aviation Products

Previous work has been done which introduced a method of detecting satellite failures and preserving integrity in an optimal fashion through the use of multiple integrated GPS/Inertial Kalman filters. The set of filters consists of a main filter that incorporates measurements from all N satellites in view and N sub-filters each excluding a different satellite. Failure detection occurs when the separation between the horizontal position solution of a sub-filter and that of the main filter exceeds a threshold that is set based upon the expected statistical separation. The horizontal protection level is computed based on this expected statistical separation as well as the covariance of each sub-filter's horizontal position error states. This method has been called the Solution Separation method of fault detection. This paper expands upon this earlier work by adding a set of sub-sub-filters for each of the sub-filters in order to perform exclusion. The logic for exclusion is presented, and a method of calculating the horizontal exclusion limit (HEL) is derived. A proposed statistical ionospheric error model is presented which accounts for the temporal and spatial correlation of the ionospheric delays for both quiet and stormy conditions. Finally, an approximate worldwide availability of both detection and exclusion is determined using this proposed ionospheric error model by first characterizing the behavior of the HPL and HEL through several RAIM holes and then applying this characterization to all RAIM holes for various satellite combinations of the optimized 24-satellite constellation.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #3

ROBUST POSITIONING WITH SINGLE FREQUENCY INERTIALLY AIDED RTK:
B.M. Scherzinger, Applanix Corporation

This paper describes an inertially aided RTK (IARTK) positioning system using single frequency receivers. The test platform is a Position and Orientation System for Land Vehicles (POS/LV) under development at Applanix, whose design objects are to achieve robust positioning with decimeter-level accuracy during various levels of satellite coverage ranging from total outage to complete coverage. The POS/LV uses a tactical grade IMU and for this experiment a single frequency receiver.

This paper first describes the IARTK architecture implemented in the POS/LV. IARTK comprises a tightly-coupled inertial/GPS integration with floated ambiguity estimation and fixed integer search in a single Kalman filter. The Kalman filter state vector contains states for inertial navigator errors, inertial sensor errors and floated ambiguities. IARTK uses the Kalman filter to tie the ambiguity search space volume to the inertial position error, which at the time of loss of lock is on the order of 2-5 centimeters. During a GPS outage, the search space volume grows at the inertial navigator position error rate, which is on the order of 10 centimeters CEP over 10 seconds and 0.5 meters CEP over 30 seconds.

The experiment comprised a series of tests of RTK performance over baselines up to 10 kilometers. The test results show that the IARTK algorithm requires between 90 seconds and 10 minutes to establish an initial L1 integer ambiguity resolution, as would be expected with a single frequency receiver. Thereafter the IARTK algorithm recovers L1 integer ambiguities within 10 seconds following outages up to 30 seconds. These results show that robust single frequency RTK positioning is possible with inertial aiding over short baselines out to 10 kilometers.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #4

INCORPORATION OF ELECTRO-OPTICAL IMAGE SENSING INTO A NAVIGATION SYSTEM:
R. Pollock, D. Sullivan, D. Boid, M. Nylund, NAVSYS Corporation

Inertial measurement can be an acceptable sole navigation source during short periods of navigation satellite signal drop-out due to intentional or unintentional interference, and provides platform attitude. However, the as the drop-out period continues, the inertial bias errors reach unacceptable levels. This problem is addressed by using relative or absolute position estimates derived from continuously captured image data to fill in for the missing satellite-based navigation capability. Both estimates are obtained by automatically localizing a model of a scene object, or landmark, in the image data. Accurate scene coordinates for the landmarks are required to generate absolute position estimates. These may be obtained from map data, or from the platform at times when accurate image sensor position and attitude estimates are available. For relative position estimates, the landmark is manually identified in one image, automatically modeled, and localized in subsequent images. This approach has been implemented and tested for aerial navigation, and the performance obtained in this situation is reported.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #5

HYBRID KALMAN FILTERING FOR GPS/DR INTEGRATION:
X. Dong, W. Li, Beijing Institute of Technology, China; HaihongWang, LiWang, S. Fan, Beijing Research Institute of Tracking and Telecommunication, China

Kalman filtering has been widely used in the GPS-based multi-sensor positioning systems. GPS integrated with Dead Reckoning (DR) is such kind of system which was found very useful in location-based tracking applications such as Personal navigation, Fleet management etc.

Kalman filtering is an estimation method that minimizes the variance of the estimation error, and assumes that the statistical properties of the noisy inputs and the dynamic model are perfectly known. This assumption, however, cannot represent the real situation in practice and thus limits the applicability of the Kalman filter. In using the GPS/DR integration system lots of traffic jam, traffic lights and high-risen buildings in urban area is the main reason that the dynamic model and statistical properties of the noise input cannot be treated as unchanged. Therefore, Kalman filtering is unable to cope with this situation. Then many robust Kalman filtering methods such as Minimax filtering have been developed, in which no knowledge of the noise statistics is assumed.

In fact, Minmax filtering, assuming too little, has the drawback of ignoring the knowledge of the noise statistics, and Kalman filtering, assuming too much, suffers from a lack of robustness. A method of combing Kalman and Minimax filtering called Hybrid filtering is used to deal with the data processing of GPS/DR integration in this paper. The analysis result and numerical examples show that the hybrid filtering performs much better than Kalman filter and Minimax filter.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #6

HIGH ACCURACY, LOW COST INTEGRATED POSITIONING SYSTEM FOR THE TORONTO TRANSIT SYSTEM:
A. El-Rabbany, A. Shaker, Ryerson Polytechnic University, Canada

Transit system authorities in many countries are faced with a challenging trend of fiscal constraints, which limits their capabilities to expand existing services and to increase the ridership. The Toronto Transit Commission (TTC) is no exception. Currently, the TTC operates nearly 1500 buses, all equipped with Dead Reckoning (DR) systems that are used for locating the buses in real-time. In addition, a total of 730 radio beacon transmitters (signposts) are placed at known locations along the bus routes. Each beacon transmits a low power microwave signal, which is detected by a receiver on the bus, to account for the DR system's drift. Unfortunately, this system has a number of limitations, including its incapability of knowing the precise location of a vehicle in between two signposts. In addition, it is not possible to track a vehicle that goes off-route as a result of, for example, a road closure.

To overcome the limitations of the current TTC system, an integrated positioning system is developed. The integrated system consists of a low cost autonomous GPS system, supplemented by the existing DR/signpost system. With the integrated system, GPS helps in controlling the drift of the DR system through frequent calibration, while the DR becomes the main positioning system during the GPS outages. In addition, the signposts are used as reference stations to account for the unmodelled GPS/DR errors. An optimal positioning solution is obtained using Kalman filtering technique, which utilizes all the available information. It is shown that, although an autonomous GPS system was used, the performance of the integrated system is comparable to that of the differential GPS (DGPS) in unobstructed environment.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #7

DEVELOPMENT AND PERFORMANCE ANALYSIS OF A TIGHTLY COUPLED GNSS/INS SYSTEM:
C. Kreye, B. Eissfeller, T. Luck, Institute of Geodesy and Navigation, University FAF Munich, Germany

Apart from space and control segment the GNSS performance depends on several parameters, e.g. signal properties, multipath conditions, signal interruptions, signal-to-noise ratio, dynamics and receiver errors like thermal noise and oscillator instabilities. The impact of these influences are based on the behaviour of the code (DLL) and carrier tracking loops (PLL) implemented in the GNSS receiver. To guarantee availability of a precise navigation solution the total loop errors have to be significantly lower as the lock thresholds. Especially under high dynamic conditions or in a jamming environment the high frequency signal dynamics have to be removed from the loops to keep them in lock.

One possibility to solve this problem is the measurement of the vehicle dynamics by an low-cost INS. Then the range velocities, accelerations and higher time derivatives can be computed by aid of inertial data and fed back directly into the GNSS receiver code and carrier tracking loops. In this way receiver phase noise and signal acquisition times can be reduced.

The paper describes theoretical and numerical analysis, system developments and tests for practical land navigation in order to evaluate sensor requirements, necessary data processing and performance of the tightly coupled system for navigation tasks.

Thus in the first part a theoretical approach is used to define error margins of INS systems used the stabilize the receiver tracking loops. The paper demonstrates that a gyro rate of approximately 1/h is necessary to aid the phase lock loop. Concerning the delay lock loop a gyro rate of 10 /h is sufficient. Additionally a possible filter design for low-cost inertial data and its transfer function are described.

In next part of the paper results of a simulation tool for a tightly coupled GPS/INS navigator are described. Algorithms and sensor models used are reviewed. A Kalman filter for estimation of the navigation states and INS errors for tightly coupling is presented. Simulation results for different observation environments and sensor errors under varying dynamical conditions are also shown. Especially the feedback effects between the GNSS and INS sensors are exposed. Comparisons between the theoretical investigations and the numerical simulations are carried out. Advantages and disadvantages of the coupled sensor system are discussed.

The development of an experimental tightly coupling system using the MITEL GPS Architect and the LITTON LN - 200 Inertial Measurement Unit is the key point of the paper. Besides hardware and software aspects methods of time synchronisation and data processing are addressed. Results for practical land navigation will show the capacity of tightly coupling also for non-military applications.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Paper #8

CALIBRATION IN MULTI-SENSOR ENVIRONMENT:
M.M.R. Mostafa, Applanix Corporation

Calibration is a critical factor in multi-sensor environment. This paper is, therefore, dedicated to present the new developments in calibrating a multi-sensor system comprising GPS, inertial measuring unit (IMU) or an Inertial Navigation System (INS), and a mapping sensor (camera, scanner, etc.). Overall calibration includes the calibration of camera/IMU boresight, the calibration of the lever arms between a camera, GPS antenna, and IMU centre, and the calibration of the imaging sensor. Two boresight calibration approaches are presented, namely the airborne and the terrestrial approach. Traditional airborne boresight calibration has been successfully used for a few years, yet it still suffers from the traditional mapping requirements such as the need for a certain number of well-distributed high precision ground control in the calibration area. The terrestrial approach, on the other hand, has been rarely used in typical map production. In this paper, the concepts of airborne and terrestrial apprpoaches of calibration are presented. The new airborne calibration approach, and the proposed terrestrial approach do not need ground control, except for quality assurance. Data results and analysis are introduced. A comparative study between the two approaches is presented for a number of data sets. The data used for analysis include a data set collected using an integrated GPS/INS/multi digital camera system; a data set collected using an integrated GPS/IMU/digital camera system, and a number of other data sets collected using integrated GPS/IMU/aerial film-camera systems. The results showed that both approaches are viable and well considerate for the operational parameters of the mapping industry.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS I
Alternate #1

DESIGN OF A LOW COST INERTIAL NAVIGATION SYSTEM WITH GPS RECEIVER FOR A CAR NAVIGATION SYSTEM:
J.S. Gil, Y.D. Cho, S.J. Park, M.P. Hyun, Elextech, Inc, Korea

This paper deals with the design of a low-cost inertial navigation system (INS) to solve the problems of GPS which is a core component of CNS (Car Navigation System). Our designed low-cost GPS/INS provides the velocity and position information of the vehicle to be tracked using very low cost three gyroscopes and one 3-axis-accelerometer.

Using this system it is possible to compensate its contained errors with the auxiliary data provided by position, velocity, heading of GPS. In order to confirm the performances of the designed INS several experiments on vehicle are conducted with GIS.

To make a hybrid GPS we designed all related H/W and S/W including GPS receiver which is so called G2000 of Elextech, Inc so that we can design and modify all components to make a hybrid GPS including the algorithms such as a kalman filter and RLS, the integration algorithms and navigation algorithms.

The used algorithm for this system is a tightly coupled one and we compared the results of the loosely coupled algorithm and the tightly coupled one.

In conclusion, the performances are improved compared to the traditional navigation system which is using only GPS receiver or GPS plus one gyroscope and a odometer added system.

Also the total cost is lower than existing CNS systems so that it is proper to adapt for land vehicle navigation.
[Back to Program]

Session F5: MULTI-SENSOR INTEGRATED NAVIGATION SYSTEMS 1
Alternate #2

GPS COMPASS : A LOW COST GPS DIRECTION SENSOR OF TWO ANTENNA TYPE:
Y. Koura, H. Suzuki, K. Ogawa, Y. Kamei, M. Nakamura, Japan Radio Co., Ltd., Japan

This paper describes a low cost GPS direction sensor of two antenna type - GPS COMPASS - which is a measuring equipment of a ship fs heading such as a gyrocompass. It consists of an antenna unit and a processing unit. The antenna unit consists of two antennas with pre-amplifiers which have been designed to have the most suitable gain and radiation pattern for accurate carrier measurements. The processing unit consists of a microprocessor, two GPS receivers and auxiliary sensors. The GPS receivers are for mobile use and very low cost. Direction angle is derived from satellite fs positions and double difference of carrier phase data output from the receivers. An integration filter is designed to combine optimally direction data derived from GPS receivers and angles measured by the auxiliary sensors. The algorithm is developed to be compact and executable on even a one-chip microprocessor. In spite of two antennas, direction, roll and pitch angles are derived from GPS receivers and auxiliary sensors, the features of this two antenna type are higher availability, easier installation and lower cost than three or more antenna types. Compared with a gyrocompass, its settling time is extremely short by using an original single epoch algorithm and it is able to output position and data speed data in addition to heading data. In order to confirm the performance and verify the accuracy and availability, many evaluation tests have been carried out in many sea trials in Japan, U.S.A. and Europe over a long period of time. This paper describes the evaluation results too.
[Back to Program]