Marine Gyrocompasses- End of 19th Century- Gyro assisted Magnetic compasses.Schuler Loop- The earth is not flat. As we move, close to the surface, we need to keep tilting the platform (with respect to inertial space) to keep the axes of the N and E accelerometers horizontal.Schuler Principle- a pendulum whose period exactly equals the orbital period of a hypothetical satellite orbiting just above the surface of the Earth (about 84 minutes) will tend to remain pointing at the center of the Earth when its support is suddenly displaced. Such a pendulum would have a length equal to the radius of the Earth.WW2- V2 guidance systems combined two gyroscopes and a lateral accelerometer with a simple analog computer to adjust the azimuth for the rocket in flight.After the defeat of Germans, They had to surrender their aircrafts and Rocket Scientists to the US who then began developing on it. They came out with Floated Rate Integrating Gyro- had drift performance of 0.01deg/hour. It was designed to be incorporated in Atlas Intercontinental Ballistic Missiles.By1980- end of cold war, the focus of developing IN for Missile technology was shifted to Space Science ( Apollo Navigation System for Space Craft) and Civil and Military Aviation ( Delta Position System and Q- refernce)
In a gimballedsystem, the gimbals can be moved intodifferent positions without removing it fromthe aircraft, thus allowing the earth'srotation and gravitational field to calibrateeach of the gyros and accelerometers. Thiscannot be done with a strapdown system. So GPS comes handy.Inflight Alignment- removing the need for the aircraft to be held stationary for up to 5 minutes while the I.N. `gyrocompasses', prior to flight.
The Kalman Filter (KF) is a very effective stochastic estimator for a large number of problems, be it in computer graphics or in navigation. It is an optimal combination, in terms of minimization of variance, between the prediction of parameters from a previous time instant and external observations at a present time instant.
Method-Processing of raw GPS Measurements through GPS Kalman Filter to determine the position and velocity from GPS.Processing of the raw INS measurements, through the mechanization eqns to determine the position and velocity from the INS.Use of Position and velocity from GPS kalman filter as input to INS Kalman Filter, which takes in the difference between the position and velocities from 1 and 2 and determines the error estimates in position and velocity and the misalignment error.Use the error estimates from it to update the position and velocity from INS mechanized values to get a full state vectors.
Method-Processing of the Raw INS measurements through Mechanization equations in order to determine position and velocity from INS.Use of Raw GPS ephemeris information and position and velocity from 1 to predict the Pseudo ranges and Doppler Measurement.Use of Predicted Pseudo ranges and Doppler Measurements from 2 and GPS raw Pseudo ranges and Doppler Measurement in order to determine the error estimates of Position and velocity and Misalignment error.Use the error estimates from 3 to update the position and velocity from 1 to get full state vectors.
Chris Hide- Research Fellow at Institute of Engg Surveying and Space Geodesy at Univ of Nottingham.Terry Moore- Director of Institute of Engg Surveying and Space Geodesy at Univ of Nottingham, Member of Council, Royal Institute of NavigationComment on point 2 of Focus of Research-Kalman Smoothening filter is generally not considered in case of Low cost INS but it can significantly improve the accuracy.
So we use Fixed-Interval Smoothening algorithm in our case as it is particularly advantageous when it is applied to GPS/INS integration during gaps in the availability of the GPS measurements.However Fixed- Lag smoothening filter can also be used for a more Real time analysis.
The figure shows that two solutions are computed in the forward and backward directions: the red line denotes the estimated error of the Figure 2 Advantage of Kalman filter smoothing during GPS outage (adapted from Gelb, 1977) forward filter estimate, and the green line denotes the estimated error of the reverse filter. The figure shows the occurrence of a data outage in the middle of the dataset.
The Rauch-Tung-Striebel (RTS) algorithm provides an efficient method for implementing Kalman filter smoothing.
A cheap GPS Reciever could have been used as we require only L1 Pseudo Range and Doppler Measurements, but cheap GPS Receiver tend to increase the noise and degrade the positioning performance.Use of High accuracy INS GPS integration system The GPS updates were only used when the carrier phase ambiguities are fixed, with the high performance IMU used to bridge any long GPS outages without significantlyreducing positioning accuracy.
Loose Coupled vs Tightly coupled-It shows that Loosely coupled INS-GPS system had poor results instead of having a better IMU ( it was because of reduced coverage of GPS satellites), But for Tightly coupled INS-GPS system, has a better result in positional solutions ( even with the use of a cheap IMU).However, it is shown that yaw error is relatively large with a standard deviation of approximately 0.9° for the tightly coupled filter. This is a common problem in low cost INS and is caused by the dynamics dependence of this state. Hence low cost tightly coupled INS-GPS system must have Kalman Smoothening Filter.The north and east velocity errors are significantly reduced by approximately 4 and 10 times respectively,2. Tightly coupled Forward only vsKalman Smoothening- It can be seen that it is an improvement over the Tightly coupled Forward INS-GPS system when Smoothening is used as it further polishes the results and the error estimates are decreased.
Figure 1-The figure shows that the solution from the forward filter suffers from relatively large positioning errors of up to 16m. But after using the smoothening it is clear that there is a significant improvement in the estimation of thevehicle position. The maximum error from the smoothed solution is 3.02m, which occurs during the longest period of reduced satellite availability.Figure 2 –shows a typical scene of the results when there is an intermediate chance of having 4-5 satellites and 4-5 building covers along the road.
Apart from the main research paper, I have gone through the following articles for understanding and creating the presentation.
AH 2916Integrated Navigation Effectiveness of INS & GPS Integration from a Urban Perspective Ipsit Dash
Outline• Technological Milestones of INS and GPS Integration• Why INS-GPS integration?• INS-GPS Integration Architectures• Research Paper• Testing and Results• Conclusions• Literature Review 2
Milestones of INS and GPS Integration• Marine Gyrocompasses- End of 19th Century• 1930-Stand alone Gyrocompasses- incorporation of Damped Schuler Loop-• 1940-WW 2- Germans (V2 Guidance Systems) and British RAE simultaneously developed IN equipments for guiding their Missiles.• 1950- Schuler Tuned IN “Floated rate Integrating Gyro” developed by MIT, USA• 1960- Devlopment of Dynamically tuned Gyro• 1970- Tremendous advancement in IN field Ring Laser Gyros ( RLG) and Hemispherical Resonator ( HRG) Strapdown mechanism used in commercial flights- Boeing 757 Nuclear Magnetic Gyros (NMR) were developed Fibre Optic Gyros (FOG)• 1980-RLG Strapdown System was used most significantly in Civil Aviation sectors. Gimballed IN systems continued to be used in Military sectors..• End of 20th Century- GPS !! A successor or a partner?? RLG Strap Down systems were improved and accuracy was improved and each unit contained a GPS receiver. 3
Why INS-GPS integration?? Most Navigation systems need to have- – Continuous and Reliable Navigation determination( Position and Orientation) – Acceptable Accuracy level and possibility of maintaining it over time GPS and INS symbiotic advantages- – Their Error Dynamics are totally different and uncorrelated. – GPS solves the problem of “calibrating” the instrument errors in a strapdown INS. – GPS provides a means of “in-flight” alignment for all INS. – The I.N. provides a seamless fill-in for GPS “outages” resulting from jamming, obscuration caused by manuvering etc. – The I.N. provides a means of smoothing the noisy velocity outputs from the GPS, and a continuous high bandwidth measurement of position and velocity. – In a tightly integrated system, the I.N. provides a means for narrowing the bandwidth of the GPS tracking loops, providing greater immunity to jamming. 4
INS-GPS integration architectures Most Common-Loosely Coupled ( Decentralized Integration) Tightly Coupled ( Centralized Integration)•2 Kalman Filters •1 Kalman Filter•Advantages- •Advantages-•Simple in application •Can be used in Urban Areas ( poor Satellite•Robustness ( Sensors aiding each other) coverage)•Small Processing time •Raw and Predicted Pseudo Range and Doppler•Disadvantage- Measurement can lead to results.•Impossible to provide measurement update •Disadvantage-from GPS Filter when GPS cover is poor. •Increase in the State Vector Sizes lead to Large processing time Other methods- Uncoupled Integration Deep/Ultra- Tight Integration •Simple Method uses GPS solution if •Uses both GPS and INS solutions to available otherwise uses INS solution update and aid each other. •Low accuracy •Requires access to GPS Receiver firmware 6
Research PaperFocus of the Research-• Use of Tightly coupled Integration for navigation in Urban Areas• Kalman Filter Smoothening algorithm to be used to post process thedata to obtain the position solution if its not available at the instant• High performance positioning can be achieved in post processing –Ideal for Low cost, High quality and Continuous Positioning. 9
Advantages of Post ProcessingApplications that can be benefitted/use Post processing to find positional solutions-• Surveying Application like Inventory Management• Terrestrial Georeferencing applications like Photogrammetric Surveying, Laser Scanning• Vehicle Performance Testing- Racing Cars or Product Testing• Surveillance- Commodity/ Vehicle Tracking• Road User- Charging where high accuracy is required and non availability of GPS service ( Urban Areas) 10
Kalman Filter Smoothing Algorithm• It is basically a post processing algorithm to find out the position solutions after all the data has been collected.• 3 basic types of Smoothening Filters-• Fixed Interval estimates the states at each of the points in a data set when all the data have been collected• Fixed Point used to estimate a specific point in a dataset• Fixed Lag can be applied in near real-time. 11
Fixed Interval Algorithm Advantage of Kalman filter smoothing during GPS outage (adapted from Gelb, 1977)When the forward and backward position estimates are computed, the INS positionerror quickly increases over time which is particularly so for systems that use a low costINS.When the Kalman filter smoothing algorithm is applied to the data, the error issignificantly reduced across the data outage interval. 12
For Near Real- Time Cases• Rauch- Tung- Striebel (RTS) algorithm can be used in near- real time by running the smoothing algorithm on short periods of data throughout during the data collection. Ex- For example, after every significant GPS outage, the algorithm could be applied once GPS availability returns.• Advantages- The RTS algorithm greatly reduces the computational effort required for Kalman filter smoothing since it only requires the full Kalman filter to be implemented in the forward direction.• Limitations- when carrier phase ambiguity states are modelled, the INS will not have the advantage of using fixed ambiguity GPS measurements that may have been resolved if a full reverse filter implementation were to have been implemented. 13
Implementation- Field Trial• A test route was devised to incorporate a range of conditions from relatively clear GPS conditions, to semi obstructed surroundings in suburban locations, to deep urban conditions where there is a significantly restricted sky view. The route was driven twice ( approx 25 mins).• Gyro Biasing Estimate- 1 min• 5 minute Alignment Period to find the Initial gyro and accelerometer biases• Initial Heading estimate was obtained from velocity of vehicle used.• Roll and Pitch initialized to 0.• IMU used- Crossbow AHRS400CA ( < $3000)• Differential GPS receiver used- Novatel OEM4 GPS (L1 Pseudo range and Doppler measurements were used. It was used with Leica GPS530 reference receiver.• The lever arm separation between the IMU and GPS antenna was calculated by using a total station.• A Novatel OEM4 GPS receiver integrated with a high accuracy Ring Laser Gyro Honeywell CIMU was also installed in the vehicle to act as a reference for the experiment. It was integrated using Loosely coupled algorithm.• The Processing was done using software – KinPos and Applanix 14
Vehicle trajectory GPS IMUs Total station survey for the estimation of lever arm separation between GPS receiver and IMUsSatellite availability during Nottingham trial 15
Kalman Filter Smoothening ResultsHorizontal position error of GPS and low Typical example of positioning errorcost INS system during Nottingham trial during restricted satellite availability 17
Conclusions• It is clear from the results discussed in this paper that GPS and low cost INS integrated systems can meet the performance levels required for a number of applications, particularly in Urban Areas.• The benefit of post-processing the data was shown to be substantial. This can be utilised in many navigational purposes with limited GPS availability. 18
Literature Review• Michael Cramer, GPS/INS Integration, University of Stuttgart- Photogrammetric Week 97- D.Fritsch & D.Hobbie, Eds 1997• Vikas Kumar N., Prof. K. Sudhakar, Integration of Inertial Navigation System and Global Positioning System Using Kalman Filtering, DEPARTMENT OF AEROSPACE ENGINEERING INDIAN INSTITUTE OF TECHNOLOGY, BOMBAY 2004• A. D. KING., Inertial Navigation – Forty Years of Evolution, Marconi Electronic Systems Ltd.• Alison K. Brown, TEST RESULTS OF A GPS/INERTIAL NAVIGATION SYSTEM USING A LOW COST MEMS IMU NAVSYS Corporation, 14960 Woodcarver Road, Colorado Springs, CO 80921 USA,• George Schmidt, INS/GPS Integration Architectures, Sponsored by the NATO Research and Technology Organization• Milan Horemuž, Integrated Navigation Compendium, Division of Geodesy and Geoinformatics, Royal Institute of Technology, Sweden 2006 19