Session B3: MEMS INERTIAL MEASURING UNITS
Paper #1

EVOLUTION OF LOW COST MEMS/GPS INERTIAL SYSTEM TECHNOLOGIES:
R.S. Anderson, D.S. Hanson, A.S. Kourepenis, Draper Laboratory

Micro Electromechanical Systems (MEMS) continue to mature as a technology and enable the demonstration of many applications not previously achievable. Advanced microfabrication techniques have produced small, low-cost silicon inertial sensors of high performance, ruggedness, and inherent symmetry. When integrated with Applications Specific Integrated Circuits (ASICs), the sensors fit in a 3-cm-per-side flat pack and operate from a single 5 Vdc supply. They have been evaluated in automotive anti-skid and traction control systems, and in guided munitions. Multi-Chip Module (MCM) versions of these instruments enable guided munitions to improve the accuracy of the Navy 5"/54 gun platform.

Additional advances in both the inertial sensor and the supporting integrated electronics and packaging technology are required to realize a low cost Inertial Measurement Unit (IMU) for applications ranging from gun launched guided munitions and guided missiles to personnel navigation and autonomous vehicles. Additional investment is also required in Global Position System (GPS) receiver packaging technology to realize an integrated MEMS/GPS INS which meets the performance, size, and cost goals of these applications.

The outcome of this investment will be Inertial Navigation Systems (INS) that are smaller (less than 3 in3), higher performing (1/hr and less than 1 mg), and lower in unit production cost (less than $1200/IMU) than achievable in any competing technology. This paper details the technology roadmap for attaining this end objective, detailing the advances achieved to date and further advances required in the sensor, electronics, and packaging arenas. Specific INS designs and requirements will be presented to better convey the complex demands and applications for this exceptional technology.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #2

UNAIDED MEMS-BASED INS APPLICATION IN A VEHICULAR ENVIRONMENT:
J. Collin, J. Kappi, J. Saarinen, Tampere University of Technology, Finland

Low-cost inertial navigation systems (INSs) have shown to be usable only for tens of seconds without additional aiding. The biggest problem is the gyro drift rate that makes the gravity removal from accelerometer measurements very difficult. This paper describes methods for extending the time between external updates when INS is used in a vehicular environment.

The objective of this research was to develop a portable INS that is applicable for different environments. That is why auxiliary velocity information from the vehicle was not wanted to use. Implemented IMU consists of MEMS accelerometers that provide relatively good accuracy, less than 700 ug rms (at 100Hz), and tuning-fork gyroscopes with 15 deg/hr rms noise. Sensor signals are used to detect surrounding environment (vehicle, pedestrian etc.) and the motion state of the IMU, and then proper algorithms are selected accordingly for each situation.

Three methods were used to reduce the INS error and thus lengthen the unaided operation time. Firstly, the effect of the acceleration error due to misinterpretation of the IMU attitude was reduced by using attitude initialization. Attitude initializations were performed when the vehicle was stationary at the traffic lights or at the crossroads so it was known that IMU was at rest. These situations were recognized by using decision function that utilizes the sensor measurements. Secondly, restrictions for INS algorithm were made so that velocity components perpendicular to the vehicle's direction of travel were removed within every INS computation cycle. Thirdly, the error growth was kept small by using a novel method for processing the input data with respect to time.

The system was tested under normal traffic conditions and performance was compared to the traditional INS solution. With additions mentioned above to the INS algorithms the system was found to be usable up to 10 minutes without aiding while the error stayed within 15 meters, 5 attitude initializations occurred. This is quite an improvement compared to the traditional INS solution using this level of sensors.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #3

DESIGN AND TEST OF LOW COST MEMS INERTIAL/GPS INTEGRATED SYSTEM FOR CAR ACCIDENT RECORDER:
Y. Jin, C.G. Park, Kwangwoon Univ., Korea

INS(Inertial Navigation System) provides very accurate attitude, velocity and position continuously. However, it is usually very expensive and its errors tend to increase with time. On the other hand, GPS(Global Positioning System) provides velocity and position information with bounded errors from signals of GPS satellites. GPS users may, however, experience short-term loss of GPS signals because of signal blockage, interference, or jamming. Because of their complementary characteristics, INS is often integrated with GPS. Recently, a low cost integrated navigation system for commercial applications such as car/personal navigation, accident record, and human body detection has been developing as MEMS gyro and accelerometer are emerging in the world.

This paper introduces the low cost INS/GPS integrated system for an accident record as well as car/personal navigation. The navigation component of the low cost INS/GPS consists of Motorola low cost GPS receiver and the low grade IMU that uses three Murata MEMS-type gyros and Sumitomo 3-axis resonant-type accelerometer. The 12bit A/D converter is used to acquire the data of inertial sensor. The 16bit microprocessor compensate sensor biases and calculate the navigation algorithm. To proceed them on real time, every tasks is managed by RTOS(Real Time Operating System), MicroC/OS-II. The main task is scheduled for each 0.01sec. For the test and application, the GUI programs are developed. They operates in PDA or desktop computer. The navigation information is monitored on PDA instantly and is verified on desktop computer by post processing.

The laboratory test and van test are carried out to verify the performance of developed system. The objective of the laboratory test is to calibrate the inertial sensors. Especially, the temperature bias, which is one of major error sources in low cost sensors, is compensated by using the temperature calibration table. The van test is mainly for checking the navigation performance on moving vehicle. Litton LP-81 midium grade IMU and Novatel Millen 3151R GPS receiver are used as the reference system. The error of integrated system is always bounded by the complementary characteristics of INS and GPS. For the performance test of accident record, the high dynamic test is performed. While the test van repeats the rapid speedup, deceleration and U-turn motion, the sensor output is recorded on flash memory and analyzed.

The analysis results show that the vehicle trajectory can be perfectly expressed on the monitor using the output of integrated system only. Therefore we can see that the low cost integrated system can be used for car accident recorder.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #4

PERSONAL POSITIONING USING WIRELESS ASSISTED GPS WITH LOW-COST INS:
H. Leppakoski, J. Saarinen, Digital and Computer Systems Laboratory, Tampere University of Technology, Finland; J. Syrjarinne, Research and Technology Access, Nokia Mobile Phones, Finland

Wireless assisted GPS (AGPS) has been considered being very promising concept for personal positioning. The network assistance has been shown to increase the availability of GPS positioning and make positioning possible also in weak signal conditions, like in office buildings or other indoor environments. Despite the improvement, there will be environments like parking garages or underground facilities where positioning is required but thick concrete walls attenuate the positioning signals below tracking thresholds.

Another drawback with AGPS as considered as means for personal positioning is that when relying on energy resources inside the handheld device only, it is also energy constrained. In these circumstances, it may be necessary to compromise between the continuous tracking requirements and duration of positioning capability.

In this research, the possibilities to overcome these weaknesses by enhancing the performance of AGPS through integration of an AGPS and low-cost INS are studied. Although the integration of GPS and INS or IMU is well known practice e.g. in navigation applications in aviation, its applicability to personal positioning requires re-examination. Low-cost INS means also lower-performance INS as measured by error growth rates. The integrated system for personal positioning may also suffer from longer, either non-intended or intended losses of GPS signals. The former case is typical when navigating behind thick walls, the latter when user prefers saving battery of the handset by relying most of the time on the MEMS-INS with lower power consumption and by allowing the more power consuming GPS receiver to operate only for short periods of time.

Special consideration in this research is paid to the interplay of the positioning accuracy and the power
consumption characteristics of GPS and INS parts of the integrated system. Different operation scenarios are studied; effect of the positioning environment to the GPS positioning accuracy is taken into account. The results will be presented as estimated positioning accuracy and maximum possible duration of continuous position tracking with single battery charge.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #5

MEMS-IMU BASED PEDESTRIAN NAVIGATOR FOR HANDHELD DEVICES:
J. Kappi, J. Collin, J. Saarinen, Digital and Computer Systems Laboratory, Tampere University of Technology, Finland; J. Syrjarinne, Research and Technology Access, Nokia Mobile Phones, Finland

Typical problem with MEMS-IMU is that the orientation information of the inertial navigator degrades rapidly with time causing large error in position estimate. Pedestrian navigation provides ways to restrict navigation solution so that the error in position estimate grows in a lower rate. The object of this work is to create pedestrian navigator using low-cost MEMS inertial sensors. This type of system can extend the operation period of the GPS integrated navigation system in unaided mode. This paper presents a new method to estimate the step-lengths and the traveled distance using low-cost MEMS inertial measurement unit. Sensor assembly consists of MEMS gyroscopes and accelerometers capable of 1 mrad/s and 0.7 mg rms noise levels, respectively, and also of magnetic compass and barometric altimeter.

The pedestrian navigator has been realized and algorithms tested using MATLAB with real sensor data. System implements footfall and motion-state recognition from inertial measurements. IMU signals are fused with magnetic compass and barometer signals by applying Kalman filter. Step-length estimations are made using acceleration measurements from single handheld IMU. There is no need for a priori knowledge of the step-length and the IMU doesn't have to be fixed in any particular orientation.

By applying this method INS operation in pedestrian navigation can be extended to several minutes without external aiding which is critical issue for handheld devices. Longer the unaided operations of the INS are more better power consumption can be achieved. Results of the navigation system are compared to standard inertial navigation system and GPS aided system approaches. The pedestrian navigator shows improvement on navigation accuracy and power consumption of the integrated navigation system in urban environment.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #6

TRIALS WITH A TIGHTLY-COUPLED GPS/INERTIAL NAVIGATION INVOLVING A MEMS IMU:
R. Pollock, D. Sullivan, K. Taylor, C. Schaffer, NAYSYS Corporation; J. Grace, Interstate Electronics

MEMS IMUs have a small size, low power consumption, and are rugged and these characteristics are important in the design of man-portable navigation systems. In addition to this, the projected prices of such units are almost an order of magnitude lower than the prices of currently available high performance FOG/SiAC IMUs. A MEMS IMU has been incorporated into a man-portable targeting system based on a laser range-finder. The IMU is the source of the range-finder line-of-sight direction. The data from this unit are tightly coupled in real-time with data from a GPS receiver in order to permit the calculation of target coordinates in real time. The usability and performance of this system, and in particular of the MEMS IMU contribution, as obtained in a series of field trials is reported.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #7

FUSION FILTER ALGORITHM ENHANCEMENTS FOR A MEMS GPS/IMU:
J.A. Rios, E. White, Crossbow Technology, Inc.

A low cost solid state GPS/IMU navigation unit has been developed that incorporates measurements from a GPS, MEMS gyros and accelerometers, and fluxgate magnetometers to provide a complete navigation solution at a high output rate. The Crossbow Technology, Inc. ADMU (Advanced Dynamic Measurement Unit) family of inertial sensor products provides stand alone solutions for vertical gyro applications, attitude and heading reference system (AHRS) applications, and full navigation GPS/IMU applications. Firmware inside the ADMU?s onboard processors produces calibrated angular rate measurements, calibrated acceleration measurements, calibrated magnetometer measurements, and the estimated navigation state which includes body attitude (roll, pitch, heading), local level horizontal navigation frame position (latitude, longitude, and altitude) and velocity at a high output rate. The algorithm used to estimate the navigation state is an Extended Kalman Filter (EKF) trajectory correction approach in which the inertial accelerometers and gyros propagate the state trajectory made up of the position, velocity, and body attitude, and the supporting sensors (GPS and magnetometers) provide ECEF position and velocity, and earth magnetic field measurements which the filter uses to calculate corrections to the trajectory state and estimate inertial sensor errors. This fusion of multiple sensors into an EKF allows for a wide variety of sensor characterizations including bias, scale factor, and unit mounting misalignment. In addition to the inertial sensor characterization, the magnetometer sensed earth magnetic field disturbances from hard-iron and soft-iron ferrous material effects are estimated and accounted for directly in the filter. Under static conditions, the attitude and heading errors are less than 0.1 degrees, and under dynamic flight tests flown against a traditional high accuracy INS system (Litton LN-100G), the attitude and heading errors are shown to be less than 0.5 degrees. The position and velocity estimates directly follow the GPS accuracy level when the GPS is providing a low GDOP solution. When the GPS accuracy level drops due to satellite occlusion, the combined solution maintains the accuracy level when compared to the INS and smoothes over GPS estimate error nonlinearities. If the GPS drops out all together, the navigation solution remains stable and will stay within an error of 1 meter in 60 seconds.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Paper #8

IMPROVING THE PERFORMANCE OF SATELLITE NAVIGATION SYSTEMS FOR LAND MOBILE APPLICATIONS THROUGH THE INTEGRATION OF MEMS INERTIAL SENSORS:
A. Kealy, P. Cross, S. Scott-Young, The University of Melbourne, Australia

The rapid growth of applications that rely on GPS has firmly established this technology as an essential component of personal, commercial and public infrastructure. Consequently, it is becoming part of our daily lives. The reality is however, that GPS does not always work. To overcome the operational failures experienced when satellite signals are obstructed, the approach adopted in international research involves integrating GPS with other sensors. In addressing the practical and conceptual limitations of this approach, the potential of modern, miniaturised inertial sensors is investigated in this research. This paper presents some practical results obtained from a system developed for land vehicle navigation that combines GPS/GLONASS measurements, with those available from a configuration of MEMS inertial sensors and the inherent intelligence of spatial information contained within a Geographical Information System (GIS). The spatial information provides additional 'measurements' that is used to constrain the navigation solution. With this approach, the solution is not dependent on the performance capabilities of the navigation sensors alone. This enables the use of lower accuracy MEMS devices, thereby reducing the cost of the navigation system while still providing a viable solution.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Alternate #1

FAA CERTIFICATION OF A MEMS ATTITUDE AND HEADING REFERENCE SYSTEM:
E. White, J.A. Rios, Crossbow Technology

A low cost MEMS inertial measurement unit (IMU) has been developed to be submitted for FAA approval. The system is a part of Crossbow Technology, Inc ADMU (Advanced Dynamic Measurement Unit) family of inertial sensor products. The system is an unaided, high performance, solid-state attitude and heading reference system intended for general aviation applications. The strap-down inertial system provides attitude and heading measurement with static and dynamic accuracy comparable to traditional spinning mass and directional gyros. The system has been designed to comply with the Federal Aviation Authority's high standards of safety and reliability, including extensive built in test (BIT) capability. Crossbow will demonstrate to the certification authority all concepts and methodologies employed to produce reliable software per guidelines in DO-178B "Software Considerations in Airborne Systems and Equipment Certification." Consideration will be given to all aspects of software production: planning, design, verification, management and quality control. Further, the system will provide an accurate inertial reference in avionic dynamic environments including altitude, temperature, shock, and vibration according to the guidelines in DO 160D "Environmental Conditions and Test Procedures for Airborne Equipment." The unit will be certified and manufactured according to the minimum performance standards described in:
* TSO-C4c - Bank and Pitch Instruments
* TSO-C3d - Turn and Slip Instrument
* TSO-C6d - Direction Instrument, Magnetic (Gyroscopically Stabilized)

In addition, a supplementary type certification (STC) will also be submitted for the unit as a primary flight instrument for Class I-III aircraft. The process of certification can be daunting, particularly the software certification effort. The paper and presentation will outline the process and describe the impact FAA specifications had on system design.
[Back to Program]

Session B3: MEMS INERTIAL MEASURING UNITS
Alternate #2

INS ERROR ESTIMATION WITH MULTI-ANTENNA GPS SYSTEM:
S. Hong, M.H. Lee, C.S. Kim, Pusan National University, Korea

GPS/INS integration has drawn much attention for the complementary properties of them. Most of the recent researches have involved with fusing GPS with relatively expensive inertial sensors. The focus was on the correction of position and velocity errors in INS with the help of GPS. An attractive new scheme of the integration is to estimate attitude error of vehicle and bias of low-cost inertial sensors with GPS measurements during navigation. With this integration scheme, a low-cost navigation system that provides relatively accurate navigation information during loss of GPS signal can be constructed.

It is shown in [1,2] that INS errors including IMU bias are observable with measurements from at least three GPS antennas for time-invariant INS error dynamics models. The error state in time-varying INS error models is observable with measurements from three antennas. Hence, using multi-antenna measurement systems, an improvement in the on-line estimation of bias of IMU as well as attitude of vehicle could be expected during navigation without applying maneuvering to the vehicle.

In this paper, characteristics of INS aided by multi-antenna GPS system are analyzed to obtain insight on the performance of the estimator for INS error. The noise statistics of IMU used in the analysis are adopted from the typical low-cost MEMS inertial sensors. It is shown that the estimation error for bias of gyros is expected to be small compared to that of accelerometers. The vertical component of accelerometer bias can be estimated much accurately than the horizontal components. The low frequency component of the estimation error for accelerometer bias is expected to be proportional to that of estimation error for attitude.

Numerical simulation results are given to verify the above observation. Performance of estimation for both IMU bias and attitude is much sensitive to the accuracy of relative position among the GPS antennas rather than to the accuracy of absolute position measurements. Based upon the simulation, a GPS measurement system desirable for the integration with low-cost IMU is discussed.

[1] Hong, S., Lee, M.H., Rios, J.A., and Speyer, J.L.,"Observability Analysis of GPS Aided INS," Proceedings of the ION GPS-2000,Salt Lake City, UT, USA, pp 2618-2624.
[2] Hong, S., Lee, M.H., Rios, J.A., and Speyer, J.L.,"Observability Analysis of GPS Aided INS," submitted to the IEEE transactions on Aerospace and Electronic Systems, November 2000.
[Back to Program]