SPAN Technology System Characteristics and Performance

Transcription

1 SPAN Technology System Characteristics and Performance NovAtel Inc. ABSTRACT The addition of inertial technology to a GPS system provides multiple benefits, including the availability of attitude output and increased position and velocity data rates. Perhaps the most valuable benefit, however, is improved solution availability in conditions where GPS alone is inadequate. When the view to GPS satellites is obstructed or intermittent, a GPS receiver is unable to provide a reliable position solution. During times when less than four satellites are available, a position cannot be calculated at all. Short outages are common in many real life applications and environments, including under forest canopies and throughout the urban canyons found in the downtown area of larger cities. Working in these conditions results in reduced productivity when relying solely on satellite technology. To improve positioning reliability in these environments, the highly stable but drift-prone observations from an inertial measurement unit (IMU) are combined with the GPS solution and/or the raw phase and pseudorange GPS observations to produce a hybrid solution. When satellite signal outages occur, a traditional GPS-only receiver will be unable to output a solution, while a GPS/INS system will use the raw IMU data to compute a solution through the outage. These complementary technologies create a system with highly accurate position, velocity, and attitude output that provides continuous positioning in challenging GPS environments. NovAtel s SPAN TM (Synchronized Position Attitude Navigation) technology provides these features by combining one of three NovAtel GPS receivers with a choice of inertial measurement units. To demonstrate SPAN Technology s performance, tests were completed to measure the position, velocity, and attitude accuracy in three environments. Data was collected while traveling under clear GPS conditions, during controlled loss of some or all satellite signals, and within an urban environment. The tests clearly highlight the benefits of adding inertial data to a GPS system using SPAN Technology. SYSTEM OVERVIEW NovAtel s SPAN (Synchronized Position Attitude Navigation) Technology seamlessly integrates GPS and inertial data for applications requiring greater functionality and reliability than traditional standalone GPS can offer. With SPAN Technology, system integrators can build the system that meets their needs by first selecting one of three NovAtel GPS receivers, each housing the OEM4-G engine: DL-4plus, with built-in memory card for data collection and integrated LCD and keypad for on-the-fly configuration ProPak-LBplus, featuring support for OmniSTAR and C correction data ProPak-Gplus, with USB capability and an RS- 3 or RS-4 interface Photos of each of the plus enclosures are shown below. Figure 1 - plus Enclosures Inertial data is added by choosing from one of two inertial measurement units, provided in NovAtel s IMU-G enclosure: IMU-G H58, containing Honeywell s HG17 AG58 inertial measurement unit (IMU) (formerly the HG17 AG11) IMU-G H6, housing Honeywell s HG17 AG6 IMU (formerly the HG17 AG17) The IMU-G enclosure is shown below. Figure IMU-G Enclosure 1 of 8 D8814 Version 1

2 With SPAN Technology, integrating the GPS receiver and inertial unit is straightforward. The IMU communicates with the receiver through one of the enclosure s standard serial ports. In the case of the DL-4plus and ProPak-Gplus, the IMU-G is powered directly from the receiver s power output. As a result, only a single cable is required from the receiver to the IMU to satisfy both communication and power requirements. For the ProPak-LBplus, a special cable has been designed to supply both the receiver and the IMU from a single power source. Along with easy to use hardware, SPAN Technology firmware combines the GPS and inertial data to provide a highly accurate solution and efficient operation. All system configuration is completed through the receiver s standard serial ports using simple commands and logs. The result is a system that is operational within minutes of installation. Features and Benefits Through the combination of GPS and inertial functionality, SPAN Technology increases productivity by offering continuous operation during periods of reduced satellite coverage. The optimized GPS/INS integration also results in faster satellite reacquisition and solution convergence. For dynamic applications requiring more than position alone, the system also provides precise velocity and attitude, with all data available at 1 Hertz for exceptional responsiveness. Improved automatic detection of stationary periods is used to the system s advantage through zero velocity updates (ZUPTS), which assist in controlling velocity errors resulting from drift in the IMU measurements. With SPAN Technology, several positioning modes are available to meet the accuracy requirements of the application. Building on the basic stand-alone mode with uncorrected GPS, more advanced positioning modes are offered for increased accuracy, including SBAS-corrected GPS,, and support for OmniSTAR and C correction services. For centimeter-level positioning accuracy, RT- mode is available. In addition, since the system is based on NovAtel s standard GPS receivers rather than custom components, integrators can easily add inertial capability to their systems after their initial receiver purchase. Existing IMU-capable receivers can be enabled to support an IMU through a quick firmware upgrade in the field. Combined with the availability of multiple receiver models and accuracy levels, this ensures that SPAN Technology can adapt and evolve as positioning requirements change. To meet the needs of applications requiring different grades of inertial performance, NovAtel continues to add popular IMU choices to its SPAN Technology product line as well. TESTING OVERVIEW The goal of the testing was to evaluate the performance of the SPAN filter across a variety of GPS conditions. Position, velocity, and attitude solution information was collected in three situations: Traveling under clear GPS conditions with no obstructions to line of sight to GPS satellites During controlled outages of some or all of the GPS signals Traveling within an urban canyon, with tall, densely located buildings blocking access to the GPS signals and introducing multipath effects Accuracy of the SPAN measurements was determined through comparison with a higher-grade control system. EQUIPMENT Testing was completed using NovAtel GPS receivers and the IMU-G enclosure housing a Honeywell inertial measurement unit (IMU). The receiver and IMU were mounted in a van and data was logged from the receiver s serial ports to a PC for storage and processing. and corrections were collected from a second GPS receiver base station. GPS Receiver and Antenna The GPS receiver under test was a NovAtel ProPak- G, containing the OEM4-G engine. A GPS-7 antenna was used. The antenna to IMU offset was measured to within two centimeters. Inertial Measurement Unit The inertial measurement unit used during testing is a tactical-grade, ring laser gyro (RLG)-based IMU manufactured by Honeywell. Specifications for the HG17 AG11 used (now ordered as the HG17 AG58) are given in the table that follows. of 8

3 Table 1 - HG17 AG11 Specifications Gyro Input Range Gyro Rate Bias Gyro Rate Scale Factor Angular Random Walk Accelerometer Range Accelerometer Linearity Accelerometer Scale Factor Accelerometer Bias Base Station Receiver and Antenna ± 1 deg/s 1. deg/hr 15 ppm.15 deg/hr ± 5 g 5 ppm 3 ppm 1. mg To provide and corrections, a ProPak-G receiver was set up on the roof of the NovAtel building with a GPS-7 antenna. The average baseline length during testing was approximately ten kilometers. Control System To evaluate the accuracy of the SPAN solution, a separate GPS/INS system with a navigation-grade Honeywell CIMU was used as the control. With a much higher accuracy than the tactical-grade HG17 and mature post-processing algorithms, the control system was shown to provide a reliable, quality solution. As a result, any discrepancy between the control and SPAN output is considered to be the error in the SPAN solution. Software Results shown are for data generated using the INS solution algorithms that run in real-time on board SPAN s OEM4-G receiver. For the majority of the testing, the data was taken from the real-time solution. In some cases, the data was processed offline to remove effects of lever arms and differential correction link outages. During offline processing, the same Kalman filter as used during real-time data collection was applied. TEST RESULTS Performance Under Good GPS Conditions To provide a baseline, testing was completed to determine the performance of SPAN Technology under clear GPS conditions. Data was collected through two test runs while driving in areas with full satellite visibility. Differential and corrections were broadcast by the base station. The steady state accuracy of the SPAN GPS/INS solution compared to the control system during single-point positioning is shown in Table. Table Steady State Performance Stand Alone Mode Test 1 Test Position Lat.59.4 Error Long (m RMS) Height Velocity Horizontal Error (m/s RMS) Vertical Attitude Roll Error, mean Pitch.. removed (deg RMS) Azimuth.4.34 The error in the steady state performance of SPAN Technology when differential corrections are applied is given in Table 3. Table 3 Steady State Performance Mode Test 1 Test Position Lat.68.8 Error Long (m RMS) Height Velocity Horizontal Error (m/s RMS) Vertical.11.1 Attitude Roll Error, mean Pitch.. removed (deg RMS) Azimuth.31.8 Application of corrections provides the error values shown in Table 4. Table 4 Steady State Performance Mode Test 1 Test Position Lat Error Long (m RMS) Height.14.1 Velocity Horizontal.1.11 Error (m/s RMS) Vertical.1.11 Attitude Roll Error, mean Pitch.. removed (deg RMS) Azimuth of 8

4 Performance During Complete GPS Outages In situations when GPS signals are unavailable, such as in tunnels, when surrounded by large buildings, or amongst heavy tree cover, traditional GPS receivers are not able to produce a solution. The SPAN GPS/INS filter can maintain accurate position, velocity, and attitude output during these conditions because IMU input to the system is not dependent on GPS satellite visibility. However, because of the nature of biases and noise on the measurements output from the IMU, the GPS/INS filter output will drift slowly with time. The rate of drift is a function of the accuracy of the IMU sensors. Very high accuracy (and very high cost) IMUs will have very slow drift rates and can coast for minutes or hours without correction. The HG17 IMU is suitable for bridging short GPS outages of less than two minutes. Solution Drift over a Single 6 Second Outage To explore the drift rate, or error, of the GPS/INS solution, GPS outages were simulated by running the real-time INS filter in an offline program. A 6 second outage during which no GPS information was available was introduced. Figure 3 shows the position error drift over this outage. Figure 3 Position Error Over a 6 Second Complete GPS Outage Start of Outage Horizontal Error Vertical Error 1 s 3 s 6 s. m.168 m.31 m.17 m.8 m.346 m Time into Test (s) After the full 6 second GPS outage, the position error has propagated to approximately.31 meters horizontally and.346 meters vertically. outage was created starting at the beginning of the valid data, that is the data generated once initialization was complete. The position, velocity and attitude errors at the end of the outage period were recorded. In each subsequent run, the GPS outage was moved forward in time by one second. When all passes of the offline program were complete, statistics were generated for position, velocity and attitude errors. The advantage of this technique is that a large number of samples is generated for each outage length. Using a traditional approach to outages, with, for example, 6-second outages separated by 6 seconds of recovery time, a 1-minute test would only yield 5 outages. Using the sweep analysis method described above, the same 1-minute test yields 6 samples (one for each second of data), which makes statistical analysis much more significant. By calculating the outage results for every part of the data set, the results will also take into account drift rates under different dynamics, for example while stopped, traveling in a straight line, or traveling around a sharp corner. The test was performed for outage lengths of 1, 3 and 6 seconds. It was performed with pseudorange differential corrections (), as well as corrections. Table 5 shows the results for the sweep test with corrections. Table 5 Solution Error Over GPS Outage Mode Outage Length 1 s 3 s 6 s Position Lat Error Long (m RMS) Height Velocity East Error North (m/s RMS) Up Attitude Roll Error Pitch (deg RMS) Azimuth corrections applied to the sweep test data result in the errors shown in Table 6. Sampled Solution Drift To further quantify the performance of the SPAN INS filter during outages, complete GPS outages with lengths of 1, 3, and 6 seconds were injected at multiple starting points in the data set collected during steady state testing. For the first run, a GPS 4 of 8

5 Table 6 Solution Error Over GPS Outage Mode Outage Length 1 s 3 s 6 s Position Lat Error Long (m RMS) Height Velocity East Error North (m/s RMS) Up Attitude Roll Error Pitch (deg RMS) Azimuth Graphically, the drift rates over time are represented in the figures that follow. Position Horizontal Error (m RMS) Position Height Error (m RMS) Velocity Horizontal Error (m/s RMS) Figure 4 Position Error Growth During Complete GPS Outages Duration of GNSS Outage (s) Figure 5 Velocity Error Growth During Complete GPS Outages Roll Error (deg RMS) Pitch Error (deg RMS) Azimuth Error (deg RMS) Figure 6 Attitude Error Growth During Complete GPS Outages Duration of GNSS Outage (s) From the data and plots, it is evident that attitude drifts the least amount over time. Attitude remains relatively stable and has drifted very little by the end of the outage period. However, errors in attitude propagate with time into velocity errors, and velocity errors propagate with time into position errors. As a result, the position error drift is the largest of the three. Performance During Partial GPS Outages During complete GPS outages, no position updates or phase updates are available for the duration of the outage. However, in reality, often there are available satellites but not enough for a position solution. In these situations, phase updates can still be applied to constrain the solution drift. One phase update requires two satellites, as the phases are differenced over time and between satellites. The inertial error drift is effectively reduced by the phase updates. The controlled outage tests were repeated with phase updates applied during the outage. These are termed partial GPS outages, since GPS positions are not used but GPS phases are. Figure 7 through Figure 9 show performance of the SPAN system during a partial GPS outage. Immediately before the outage, the system was operating in mode. Velocity Height Error (m/s RMS) Duration of GNSS Outage (s) 5 of 8

6 Figure 7 Position Error Growth During Partial GPS Outages Position Horizontal Position Height Phase 1 Phase Phase 3 Phase Duration of Partial GNSS Outage (s) Figure 8 Velocity Error Growth During Partial GPS Outages Velocity Horizontal Phase 1 Phase Phase 3 Phase While the analysis from controlled outages does provide some insight into the best-case operation of SPAN Technology during limited satellite visibility, it is not an accurate reflection of conditions typically encountered by users. The controlled outages are clean, as the satellites disappear instantaneously, not one by one as they might under actual operation with the GPS position degrading as they go. There is also little multipath during the test, whereas this would likely not be the case in conditions where the GPS tends to be blocked, such as near high buildings. Performance in Urban Canyon To measure the performance of SPAN Technology in a more realistic setting, data was collected while driving through downtown Calgary, Alberta, Canada. The downtown area is densely populated with tall buildings blocking access to GPS satellites. Partial or full outages are frequent and multipath effects are high. Figure 1 provides a view of one of the streets traveled during testing. Figure 1- Downtown Calgary, Alberta, Canada Velocity Height Duration of Partial GNSS Outage (s) Figure 9 Attitude Error Growth During Partial GPS Outages Roll Error (deg) Pitch Error (deg) Azimuth Error (deg) Phase 1 Phase Phase 3 Phase Duration of Partial GNSS Outage (s) As shown in the plots, when a limited number of satellites are available, phase updates can be used to constrain the growth of errors in the solution. As the number of satellites available increases, with two available satellites corresponding to one phase update, the growth in the error is constrained further. The results from one downtown test run are presented in detail below. Errors are defined in comparison to the same inertial control system employed in the clear sky testing. Position Error Figure 11 shows the trajectory traveled during the test as calculated by both the GPS/INS filter and the filter alone, without inertial aid. The plot shows large sections with no GPS solution and highlights the noisy nature of the GPS-only solution, as the solution is attempted under high multipath conditions with a frequently changing and often insufficient constellation. SPAN provides a continuous trajectory, as well as identifying and rejecting poor quality GPS solutions. 6 of 8

7 North Displacement (m) Figure 11 Plot of Calculated Trajectory in Urban Canyon East Displacement (m) GNSS/INS Figure 1 shows a zoomed-in section of one of the more difficult streets. The GPS/INS solution is clearly more accurate and reliable than the position. North Displacement (m) Figure 1 Enlarged Section of Trajectory GNSS/INS East Displacement (m) The blended GPS/INS solution available from SPAN has a much better solution error than the GPS-only solution. Table 7 that follows shows the quantified position error during the test for both the GPS and GPS/INS solutions, as compared to the control system. Table 7 Position Accuracy in Urban Canyon Position Horizontal Vertical RMS Max RMS Max (GPS-Only) /INS (GPS-Only) /INS To take a closer look at the SPAN results in mode, the position errors over time along the trajectory are shown in Figure 13. Figure 13 - Position Error in Urban Canyon over Time D Position Height Sigma Envelope x GPS Time (s) x 1 5 The standard deviation reported by the GPS/INS filter was used to compute the 3-sigma envelope shown on the plot, while the errors are computed with respect to the control system. The filter statistics accurately portray the true error of the system. Solution Availability Due to shadowing by tall builds in the region of the test, the availability of a GPS-only solution during the test was limited. Table 8 below shows the solution availability for the GPS-only and GPS/INS filters, confirming that the addition of inertial capabilities results in more reliable positioning in these conditions. Table 8 Solution Availability in Urban Canyon Total One Second Solution Epochs Percentage of Possible Solution Epochs (GPS-Only) % (GPS-Only) 151 6% GPS/INS 49 1% Under the conditions of this test, the GPS-only systems were only able to compute valid solutions between 6 and 8 percent of the time. The SPAN combined GPS/INS solution gave 1 percent availability through the urban canyon conditions. 7 of 8

8 Reacquisition SPAN also greatly improves GPS signal tracking reacquisition and solution availability. After the restoration of a full constellation following an outage, SPAN can reacquire a fixed narrowlane solution in just over 1 seconds, 9% of the time. This is an improvement of 78% over GPS alone, which takes approximately 56 seconds, 9% of the time, under the same conditions. The 95 th percentile of L1 GPS signal reacquisition is improved from 11.1 s to 1.4 s when running SPAN. Figure 14 shows the cumulative histogram of L1 signal reacquisition when testing a GPS-only OEM4- G receiver against an OEM4-G receiver running SPAN. Figure 14 - L1 Signal Reacquisition Historgram SUMMARY As shown in the test results, the addition of inertial functionality to a GPS system results in improved data availability and reliable operation in conditions where GPS alone is inadequate. During times of no satellite visibility, SPAN Technology is able to continue providing position, velocity, and attitude data. However, the biases and noise on the measurements output of the IMU cause the GPS/INS solution to drift slowly over time. During partial coverage, this drift rate can be controlled through the application of phase updates for even greater accuracy over an extended outage. In real world conditions, such as Calgary, Alberta s urban canyon, SPAN Technology provides more reliable solution output and assists with identifying and rejecting poor quality GPS solutions for improved accuracy over GPS-only systems. In addition, SPAN Technology improves GPS signal tracking reacquisition and solution availability. This level of performance, combined with ease of integration, makes SPAN Technology an ideal system for integrators looking to increase their efficiency in difficult GPS conditions while benefiting from high data output rates. For more information on SPAN Technology, contact NovAtel at 1-8-NovAtel or or visit our website at 8 of 8