Japan-based Nikken Lease has tapped into positioning and cellular technologies from u-blox for its new trackable pallet, Transeeker.
Transeeker is a pallet equipped with u-blox’s cellular and positioning technologies for accurate tracking: u-blox cellular UMTS/HSPA(+) module LISA-U200-62S and u-blox 7 standalone GNSS module, EVA-7M.
For the first time, Nikken Lease is offering its GNSS-enabled pallets as a rental service. Made of plastic, the pallets are reusable, making them a good ecological alternative to their wooden counterpart.
“When we designed Transeeker, we wanted to ensure that we could locate a pallet precisely and retrieve it nationwide at low costs for our customers,” said Tsumura, planning director for Nikken Lease. “u-blox is just the right partner with excellent product performance and unique features such as CellLocate.”
Embedded in Transeeker, the LISA-U200-62S of the LISA-U2 series is equipped with CellLocate, u-blox’s proprietary hybrid positioning technology enabling stand-alone location estimation based on surrounding GSM/UMTS cell information in conjunction with GPS positioning data.
The module also offers worldwide W CDMA(UMTS) and GPRS/EDGE coverage, and an easy migration to u-blox GSM/GPRS, CDMA and LTE modules. Also found in Transeeker, the EVA-7M single GNSS module features the reliable performance of the u-blox 7 positioning engine (receiving GPS, GLONASS, QZSS and SBAS signals) and delivers high sensitivity and minimal acquisition times in the ultra-compact EVA form factor.
“This collaboration with Nikken Lease should help us strengthen our position in the Japanese market as a key player for IoT applications,” explains Tesshu Naka, Country Manager of u-blox Japan. “We are looking forward to more collaboration with Nikken Lease.”
Geodata is key to the digital future and a 4.0 business world, according to a new report released at InterGeo in Hamburg, Germany. At the heart of this business vision is the networking of sensors that must have location data in order to fulfill their value.
The 116-page Intergeo Report, in parallel German and English, includes sections on smart cities, public participation, autonomous driving with live mapping, and surveying on the open seas. An eight-page GNSS Update section features CEOs answering questions market focus of their GNSS products, the role of geo-referencing in the Internet of Things, the coming-of-age of precise point positioning (PPP), and the opportunities for GNSS opened up by autonomous driving.
Access to company-specific geodata offers managers in the automotive industry a competitive ad- vantage. Apps show today’s motorists the way to the nearest electrical charging station. Soon, the same motorists will talk to their on-board computer to find a parking space. It will guide them instantly to the nearest free space. Geoinformation will then no longer just be found in the satnav but also in the integrated sensor in the road paving infrastructure and in the status reports of other road users.
Networking Everything. The Internet of Things is taking shape and permeating all areas of life. At its center are the tiny pieces of information that assign coordinates to a parking space, a loading berth for a container ship, a screw in the shelves of a supplier’s warehouse, or the alarm system of a family home. Degrees, minutes and seconds show people the way, answer a range of questions and help make informed decisions. Geoinformation is both an asset and an essential source of information.
Content Is King. Key companies in the geoinformation sector have naturally taken onboard the value of geoinformation. It forms the basis of their business activities. The use of geodata as added value for their products is still very new. Esri realized early in the sector that selling software is no longer sufficient on its own. Only data enables customers to harness the value of products. Cloud solutions store the mountains of data, while platforms deliver the answers.
Such new business leading lights as AirBnB, Uber, Facebook and Google could not survive without geoinformation. It is part of increasingly intelligent systems that make users’ lives a little easier and more comfortable, optimizing processes and enabling people to operate and participate in ways that were previously impractical or impossible.
The examples are myriad. Consider just a few. Digitally aided planning and construction in building information modeling not only streamlines processes and reduces costs, it enables public participation in planning procedures, using digital models of planned reality. Aerial surveys and data gathering by UAV, not only for traditional survey needs but for growing requirements in natural resource planning and management, infrastructure inspection and maintenance, surveillance and security, and more. Guidance systems for the blind.
All require location data. GNSS (satnav) is the core supplier of this data, but must be augmented by other technologies in special environments.
Releasing Geodata Pays Dividends. Managers of geodata realize they need to release it in order for it to lead them to “more” – more value, more benefits, more transparency, more importance. Geoinformation and digitization are inextricably interlinked, and this is just the beginning.
SBG Systems displays their full range of MEMS-based inertial sensors at InterGeo 2016, with a major firmware update for its Ekinox and Apogee product lines. The key improvements in the update include a 15% improvement on orientation and navigation data and better robustness under harsh environments. This firmware is a complete rework of existing functionalities with the addition of new features and improved configuration interface to ease device configuration.
Performance. Up to 15% inertial navigation system (INS) performance improvement from a reworked data fusion algorithms; and improved performance using NMEA GNSS aiding.
Ease of use. Alignment and new status flags have been added to ensure the unit reaches optimal accuracy. The unit can now compute and output on each port a full deported navigation and ship motion data. A completely reworked web interface with 3D views eases mechanical installation. Stability and reliability improvements are reported, especially while using two GNSS at the same time
Various input and output protocols have been added. See SBG Systems website for further information.
GPS World staff is reporting from Intergeo Oct. 11-13 in Hamburg, Germany. The massive trade show is considered the world’s leading conference trade fair for geodesy, geoinformation and land management. With more than 16,000 visitors from 80 countries, it is one of the key platforms for industry dialogue.
Swift Navigation has announced its newest product, Piksi Multi, a multi-band, multi-constellation high-precision GNSS receiver for the mass market.
A San Francisco-based startup, Swift Navigation introduced the first Piksi GNSS receiver in January.
Swift Navigation will be showing Piksi Multi at InterGeo Oct. 11-13 in Hamburg, Germany. The company’s booth is located in Hall A1, in the US Pavilion, booth #B1.061.
Autonomous devices require precision navigation, especially those that perform critical functions. Swift Navigation solutions use real-time kinematics (RTK) technology, providing location solutions that are 100 times more accurate than traditional GPS.
Piksi Multi supports GPS L1/L2 and is hardware-ready for GLONASS G1/G2, BeiDou B1/B2, Galileo E1/E5b, QZSS L1/L2 and SBAS. Multiple signal bands enable convergence times measured in seconds, not minutes. Multiple satellite constellations enhance availability in new environments.
The Piksi Multi with an evaluation board.
The Piksi Multi Evaluation Kit also has been upgraded with all-new components. The new kit contains two Piksi Multi GNSS modules, two integrator-friendly evaluation boards, two GNSS survey-grade antennas, two high-performance radios, so that it can deliver best-in-class reliability and range — well over 10 kilometers — and all of the accessories required for rapid prototyping and integration.
Swift Navigation expects Piksi Multi to ship in early in the first quarter of 2017. The company is accepting pre-orders in its online store at www.swiftnav.com.
Piksi Multi is an open platform. It enables customers to run Linux OS on its second core, allowing them to quickly prototype and adopt their own applications in a well-known and widely used environment.
Industries standing to benefit most from the new product include: autonomous vehicles, UAV, precision agriculture, robotics, space, survey and control and R&D applications requiring precise positioning.
Swift Navigation was built on the notion that highly-precise RTK solutions should be offered at an affordable price. Benefits of Piksi Multi for customers include:
Centimeter-level accuracy using RTK
Fast convergence times using multi-band
Robust positioning using onboard MEMS hardware
Open platform with onboard Linux
Rapid prototyping with a complete evaluation kit
Future-proof hardware with in-field software upgrades
“With the launch of Piksi Multi, Swift is taking another huge step forward in delivering affordable and highly-precise GNSS technology,” said Swift Navigation CEO, Timothy Harris. “Piksi Multi will continue to revolutionize the autonomous devices category, which is growing at an unbelievable rate.”
The United States Naval Observatory (USNO) has selected NovAtel’s GPS Anti-Jam Technology (GAJT) to satisfy a requirement for a controlled reception pattern antenna capability at sites throughout the Department of Defense Information Network (DoDIN).
The DoDIN is the core global enterprise network of the United States military and is depended upon for secure and sensitive voice, data, video and bandwidth services. This latest order brings the number of NovAtel GAJT antennas ordered by the U.S. Navy to more than 600.
GAJT protects GPS-based navigation and precise timing receivers from intentional jamming and accidental interference. It is a null-forming antenna system that ensures satellite signals necessary to compute position and time are always available.
The commercial off-the-shelf product comes in versions suitable for land, sea, fixed installations and smaller platforms such as UAVs. Military vehicles and platforms, networks and timing infrastructure also benefit from the protection that GAJT provides. There is no need to replace GPS receivers already installed, as GAJT works with civil and military receivers, and is ready for M-code, according to NovAtel.
NovAtel’s manufacturing techniques and quality processes mean that that the company can ramp up quickly to meet volume requirements, the company said.
“This order underlines our ability to deliver GAJT in volume and on time,” said Michael Ritter, president and CEO of the Canada-based NovAtel. “GAJT has now been shipped and is in use operationally by 12 allied nations around the globe. We are grateful for the rigorous technology selection process conducted by USNO which led to this latest order.”
The U.S. Naval Observatory is located in Washington, D.C.
Located in Washington, D.C., the USNO is one of the oldest scientific agencies in the United States, with a primary mission to produce Positioning, Navigation and Timing for the United States Navy and the United States Department of Defense.
The U.S. Department of Transportation will host its fifth workshop on the GPS Adjacent Band Compatibility Assessment effort on Oct. 14 in Washington, D.C. The workshop is open to the general public by registration only. Those who would like to attend the workshop are asked to register by Tuesday, Oct. 11.
The purpose of this workshop is to discuss the results from testing of various categories of GPS/GNSS receivers to include aviation (non-certified), cellular, general location/navigation, high precision and networks, timing, and space-based receivers. The workshop also will include a discussion on the development of use-case scenarios for these categories.
By Sashidharan Manickam and Kyle O’Keefe PLAN Group, Department of Geomatics Engineering, University of Calgary
This paper analyzes the GNSS signal authentication limits in using different grades of IMU (Tactical and MEMS) to detect errors in combination with different grades of GNSS receiver (Geodetic grade and Automotive). To test these combinations, a tightly-coupled 23 state navigation Kalman Filter is implemented with a constant velocity dynamics model for the position, velocity, attitude and clock states and first-order Gauss-Markov processes to model the 12 sensor errors.
These columns have focused on procedures and routines for establishing GNSS-derived orthometric heights. There are many ways to analyze and investigate GNSS data and adjustment results. I have provided some basic concepts that I believe are important for users to understand.
The selection of constraints is a very important part of establishing accurate and consistent NAVD 88 GNSS-derived orthometric heights. All of the analysis and recommendations have been based on using the National Geodetic Survey‘s latest scientific geoid model.
I recommend first performing the analysis using the scientific geoid model because the hybrid geoid model has been warped to be consistent with the published NAVD 88 values. However, as mentioned in Part 7 (June 2016), in practice, GNSS-derived orthometric heights are incorporated into the NAVD 88 using the latest hybrid geoid model GEOID12B. This column will focus on the NGS “GPS on BMS (GPSBM)” dataset that was used to create the hybrid geoid model.
As mentioned in Part 3 (October 2015), the hybrid geoid model is designed to fit the published NAVD 88 leveling-derived orthometric heights. Saying that, the GPSBM dataset can be used to identify potential issues in the NAVD 88 published orthometric heights. GNSS users should be familiar with this dataset and how it can be used in their analysis. This column will provide tools and routines that can be used to identify potential issues in NAVD 88 heights and/or NAD83 (2011) published ellipsoid heights.
Each of the below regions uses variants of the NAD 83 reference frame and a local vertical datum. Several versions of NAD 83 exist conforming to significant plates: Pacific, Mariana, and North America. Likewise, each region has its own vertical datum. It is not possible to level across water, so islands will have selected a tide gauge to serve as the local datum point and all leveling is tied to that site. The only exception to this is Hawaii. No tide gauge was selected in the Hawaiian Islands and no vertical datum has been established as of yet. Hence, GEOID12B in Hawaii transforms between NAD 83 (PA11) and the same geopotential (geoid) surface as the USGG2012 model ( W0 = 62636856.00 m**2/s**2).
Items that are listed in the below table include the final GPSBM files for each region as both Excel spreadsheets and text files as well as thumbnail images linked to larger images showing the distribution of the GPSBM’s. Alaska and the island regions are more consistent, so not many points were dropped and each is provided in its own spreadsheet/text file and identified with the appropriate ellipsoidal reference frame and level datum (see below).
The most significant work occurred in the COnterminous United States (CONUS). For CONUS, there were 24,782 points with 911 rejected leaving 23,961. These were supplemented from the OPUS-database with 737 points of which 238 were rejected leaving 499. There were also 579 points in Canada with 5 rejected leaving 574. In Mexico, there 744 of which 497 were clipped since they were too far south and another 70 were rejected leaving 177. This brings a total of 26,932 points of which 1,721 were rejected or clipped and 25,211 retained for modeling GEOID12B. The data in Canada and Mexico provide continuity up to and across the U.S. borders but do not make the GEOID12B model valid in those countries.
Points were rejected either because the State Advisor recommended it be dropped (e.g., known subsidence region), the residual ellipsoid height errors (from the NA2011 project) indicated a point was too noisy in comparison to other points in a state/region, the orthometric height was suspect, or the residual errors during geoid modeling were too high. The corresponding error flags are ‘S’, ‘h’, ‘H’, and ‘N’ as seen on the spreadsheet and text files. These points then represent the control data that were used to define the transformation between NAD 83 and NAVD 88 for CONUS.
The control data were much simpler in other regions due to the lack of quantity (more than two orders of magnitude less). Data in these regions follows a similar pattern where some data are rejected based on the codes given above for CONUS. The columns on the right side give the respective datums realized by GEOID12B for each region.
Table 1 is an excerpt of the excel spreadsheet for the GPSBM dataset and provides a sample of the contents. The headings of the columns are fairly self-explanatory. What’s important here is that the excel spreadsheet provides the name, latitude, longitude, NGS’ PID, the ellipsoid height and orthometric height of the stations used in making GEOID12B.
Table 1
Excerpt of the Excel spreadsheet for GPS on benchmarks (GPSBM) used to make GEOID12B.
The “GPS On Bench Marks (GPSBM) Used To Make GEOID12B” write up states that 1,721 stations were rejected and were not used in developing the hybrid geoid model. It also states that for the conterminous United States (CONUS), there were 24,782 stations with 911 rejected leaving 23,961. This column is going to focus on CONUS but the analysis can be performed everywhere.
As the write up states, stations were rejected for four different reasons:
Code h – The residual ellipsoid height errors from the NAD 83 (2011) project indicated that the point was too noisy,
Code H – The orthometric height was suspect,
Code N – The residual errors during geoid modeling were too high.
These rejected stations were not used to make the hybrid geoid model but since the hybrid geoid model is distorted to fit the NAVD 88, these rejected stations as well as stations nearby the rejected stations should be re-evaluated using the latest scientific geoid model, e.g. xGeoid16b.
So, what should the user do with the GPSBM table? I recommend that users perform the following steps when analyzing the stations in the GPSBM table.
Step 1: Compare the modeled GEOID12B (N12B) value to the computed GPS/Leveling (h minus H) value using the following formula: Published N12B from the NGS data sheet minus (ellipsoid height from the GPSBM table minus orthometric height from the GPSBM table). We discussed this procedure a year ago in Part 3 (October 2015). It should be noted that the orthometric height in the GPSBM table may be different than the published NAVD 88 height on the NGS data sheet if the station has been readjusted since the GPSBM table was created.
Step 2: Repeat the procedure in Step 1 using the latest NGS experimental geoid model, e.g. xGeoid16b. At this time, NGS only provides the experimental geoid models referenced to IGS08 so the user will have to use NGS’ xGeoid16 web tool to obtain the station’s IGS08 ellipsoid height and xGeoid16b value. The input to the tool is the station’s NAD 83 (2011) coordinates (latitude, Longitude, and ellipsoid height). [An example of using the xGeoid16 web tool is provided in the box titled “Example of Using NGS xGeoid16 Web Tool.”] As discussed in Part 3 (October 2015), the user will have to remove a bias and trend based on the differences in the region.
The user could also transform xGeoid16b/IGS08 geoid values to xGeoid16b/NAD 83 (2011) geoid values using their own tools, and then remove a bias and trend based on the differences. Michael Dennis, a PhD candidate at Oregon State University, created an ArcGIS raster of the xGeoid16b model, where his model has been referenced to NAD 83 (Michael L. Dennis, RLS, PE, MS Civil Eng., Geodetic Analysis, LLC, 55 Creek Rock Road, Sedona, AZ 86351). He removed a trend using the GPS/Leveling data set as input; therefore, this raster file is a form of a hybrid geoid model distorted only to remove the tilt assumed to be in the NAVD 88. I will refer to this model as Geoid16B_NAD83 to avoid confusion with NGS’ xGeoid16b model.
*Orthometric height difference between xGEOID16B to model shown
Step 3: Use the station’s data sheet to identify how the station’s orthometric height was determined; for example, was it rigorously adjusted into the NAVD 88 (published height attribute – Adjusted). We discussed the attributes of the NGS data sheet in Part 5 (February 2016). A summary of the attributes from the NGS data sheet DSDATA.TXT file is provided in the box titled “Extracted from NGS’ DSDATA.TXT.” I have highlighted the most common attributes of the stations involved in making GEOID12B.
Extracted from NGS’ DSDATA.TXT
***************************************************************************
* dsdata.txt *
***************************************************************************
There are various Vertical Control sources, as specified below:ADJUSTED = Direct Digital Output from Least Squares Adjustment of Precise Leveling.
(Rounded to 3 decimal places.)ADJ UNCH = Manually Entered (and NOT verified) Output of Least Squares Adjustment of Precise Leveling.
(Rounded to 3 decimal places.)
POSTED = Pre-1991 Precise Leveling Adjusted to the NAVD 88 Network After Completion of the NAVD 88 General Adjustment of 1991.
(Rounded to 3 decimal places.)
READJUST = Precise Leveling Readjusted as Required by Crustal Motion or Other Cause.
(Rounded to 2 decimal places.)
N HEIGHT = Computed from Precise Leveling Connected at Only One Published Bench Mark.
(Rounded to 2 decimal places.)
RESET = Reset Computation of Precise Leveling.
(Rounded to 2 decimal places.)
COMPUTED = Computed from Precise Leveling Using Non-rigorous Adjustment Technique.
(Rounded to 2 decimal places.)
GPSCONLV = Leveled Orthometric Height tied to GPS HT_MOD Orthometric Height.
(Rounded to 2 decimal places.)
LEVELING = Precise Leveling Performed by Horizontal Field Party.
(Rounded to 2 decimal places.)
H LEVEL = Level between control points not connected to bench mark.
(Rounded to 1 decimal places.)
GPS OBS = Computed from GPS Observations.
(Rounded to 1 decimal places.)
VERT ANG = Computed from Vertical Angle Observations.
(Rounded to 1 decimal place; If No Check, to 0 decimal places.)
SCALED = Scaled from a Topographic Map.
(Rounded to 0 decimal places.)
U HEIGHT = Unvalidated height from precise leveling connected at only one NSRS point.
(Rounded to 2 decimal places.)
VERTCON = The NAVD 88 height was computed by applying the VERTCON shift value to the NGVD 29 height.
(Rounded to 0 decimal places.)
Step 4: Use the station’s NGS data sheet to determine the adjustment date of the station’s published NAVD 88 orthometric height. We discussed this in Part 7 (June 2016). As mentioned in Part 7, if the station has a different adjustment date than other stations nearby, there could be inconsistencies due to adjustment distribution corrections and/or movement.
Step 1 was demonstrated in Part 3 (October 2015) so we don’t need to describe the process in this column. Comparing published GEOID12B values with computed values is the first step; the difference is an indication of how well the data fit the model and can be useful for identifying large outliers. It can be helpful in prioritizing where additional observation should be obtained when there are limited resources. Provided below is an example of where to obtain the information for comparing the modeled GEOID12B (N12B) value to the computed GPS/Leveling (h minus H) value using the following formula: Published N12B from the NGS data sheet minus (ellipsoid height from the GPSBM table minus orthometric height from the GPSBM table). The user can obtain the GEOID12B value from the NGS data sheet [see box titled “Excerpt from NGS Data Sheet For Station L 275 (HW2088)”]; for this example, the GEOID12B value for station L 275 is -30.813 m. Table 2 is an excerpt from the GPSBM file that contains the ellipsoid height (599.253 m) and the orthometric height (630.016 m) for station L 275. It should be noted that the ellipsoid and orthometric heights in the GPSBM table are given in millimeters. The first row of table 3 provides the results of the computation: [-30814 mm – (599253 mm – 630016m m) = 51 mm], or 5.1 cm.
Table 2
Excerpt of the Excel spreadsheet for GPS on benchmarks (GPSBM) used to make GEOID12B – Stations on plots in this column.
Excerpt from NGS Data Sheet For Station L 275 (HW2088)
PROGRAM = datasheet95, VERSION = 8.9.1
1 National Geodetic Survey, Retrieval Date = OCTOBER 1, 2016
HW2088 ***********************************************************************
HW2088 CBN – This is a Cooperative Base Network Control Station.
HW2088 DESIGNATION – L 275
HW2088 PID – HW2088
HW2088 STATE/COUNTY- WV/RANDOLPH
HW2088 COUNTRY – US
HW2088 USGS QUAD – MILL CREEK (1995)
HW2088
HW2088 *CURRENT SURVEY CONTROL
HW2088 ______________________________________________________________________
HW2088* NAD 83(2011) POSITION- 38 43 54.95105(N) 079 58 19.75931(W) ADJUSTED
HW2088* NAD 83(2011) ELLIP HT- 599.253 (meters) (06/27/12) ADJUSTED
HW2088* NAD 83(2011) EPOCH – 2010.00
HW2088* NAVD 88 ORTHO HEIGHT – 630.016 (meters) 2066.98 (feet) ADJUSTED
HW2088 ______________________________________________________________________
HW2088 NAD 83(2011) X – 867,581.099 (meters) COMP
HW2088 NAD 83(2011) Y – -4,906,352.726 (meters) COMP
HW2088 NAD 83(2011) Z – 3,969,521.039 (meters) COMP
HW2088 LAPLACE CORR – 0.13 (seconds) DEFLEC12B
HW2088 GEOID HEIGHT – -30.814 (meters) GEOID12B
HW2088 DYNAMIC HEIGHT – 629.553 (meters) 2065.46 (feet) COMP
HW2088 MODELED GRAVITY – 979,873.5 (mgal) NAVD 88
HW2088
HW2088 VERT ORDER – FIRST CLASS II
HW2088
HW2088 Network accuracy estimates per FGDC Geospatial Positioning Accuracy
HW2088 Standards:
HW2088 FGDC (95% conf, cm) Standard deviation (cm) CorrNE
HW2088 Horiz Ellip SD_N SD_E SD_h (unitless)
HW2088 ——————————————————————-
HW2088 NETWORK 1.00 1.94 0.45 0.36 0.99 -0.05669181
Table 3 contains the comparisons between modeled geoid values and their computed geoid values for five station pairs that have large relative differences. Looking at table 3 one can see that there are several large relative differences between the published GEOID12B model and computed geoid model (see column titled “N12B minus (h-H)” in table 3). This doesn’t mean that the model is incorrect, it only means that there were large relative differences that the model had to account for. As previously mentioned, GEOID12B was created to be consistent with the NAVD 88.
Since the experimental geoid model xGeoid16b_NAD is not distorted to conform to the NAVD 88 everywhere, it should provide better information for identifying outliers and determining which stations appear to be inconsistent with its neighbors.
Figure 1 – All GPS on BMS Residuals Using Geoid16b_NAD model (note: rejections by geoid team have been removed).
Table 3
Table of selected stations involving large relative differences depicted in plots in this column.
(Results are provided for GEOID12B and Geoid16B_NAD Models*) *Michael Dennis, a Ph.D. candidate at Oregon State University, created the xGEOID16B ArcGIS raster, where the model has been referenced to NAD 83 with a trend and bias added to account for the apparent tilt in the NAVD 88. This model is denoted as Geoid16B_NAD (N16b) in this column.
Figure 1 is a plot of all of the GPSBM residuals using the Geoid16B_NAD83 model. This plot indicates that there are a lot of large residuals. First, let’s define what I’m calling residuals. The residuals on my plots are the differences between the modeled geoid height value and the computed geoid height value using the ellipsoid height (h) and orthometric height (H) from the GPSBM data set; that is, residual = modeled gravity value – (h minus H). The largest negative residual is -37.3 cm and the largest positive residual is 33.8 cm.
Figure 2 – Positive GPS on BMS Residuals Using Geoid16b_NAD model (note: rejections by geoid team have been removed).
Figure 2 is a plot of the positive GPS on BMS residuals using Geoid16b_NAD geoid model. There are 5957 residuals greater than 5 cm (not including the stations rejected by the NGS geoid team). As you can see, it appears that most of the positive residuals are on the eastern half of the United States.
Figure 3 – Negative GPS on BMS Residuals Using Geoid16b_NAD model (note: rejections by geoid team have been removed).
Figure 3 is a plot of the negative GPS on BMS residuals using Geoid16b_NAD geoid model. There are 4113 residuals less than -5 cm (not including the stations rejected by the NGS geoid team). As you can see from the plot, the negative residuals appear to be more evenly distributed across the United States than the positive residuals. It does, however, appear that there are more negative residuals greater than -5 cm along the Gulf Coast, Atlantic Coast, and the Great Lakes than there are positive residuals greater than 5 cm. In addition, there appears to be a lot of negative residuals in the northeastern United States.
Figure 4 – GPS on BMS Residuals Using Geoid16b_NAD model in North Carolina and South Carolina (note: rejections by geoid team have been removed).
Figure 4 is a plot of the GPS on BMS residuals using the Geoid16b_NAD geoid model in the North Carolina and South Carolina border region. What’s interesting about this plot is that South Carolina doesn’t seem to have many negative residuals where North Carolina has both negative and positive residuals. We will look at this in more detail later in this column.
Figure 5 – GPS on BMS Residuals Using Geoid16b_NAD model in Washington and Oregon Region (note: rejections by geoid team have been removed).
Figure 5 is a plot of the GPS on BMS residuals using Geoid16b_NAD model in the Washington and Oregon Region. This graphic shows some large grouping of negative and positive residuals, especially along the Pacific Coast in Northwestern Washington State.
Now, let’s look at some large relative differences in residuals between stations that are spatially close together. Figure 6 is a plot of large relative differences between groups of GPS on BMS residuals (using Geoid16b_NAD model) at the North Carolina/South Carolina border. In figure 6, two stations (FA1337 and FA1560) are about 20 km apart and the difference in residuals is -18.6 cm (-12.4 cm minus 6.2 cm). This is a large difference for only 20 km. What is even more significant is that the group of stations near FA1337 are all negative residuals (around -10 cm) and the group of stations near FA1560 are all positive residuals (around 6 cm), this could be an indication of a large distribution correction due to the NAVD 88 design. We discussed the distribution correction in Part 7 (June 2016). These stations definitely needs to be investigated.
The next step in my process is to look at the NGS data sheets for these stations to determine how the stations were adjusted.
Step 3: Look at the station’s data sheet to identify how the station’s orthometric height was determined; for example, was it rigorously adjusted into the NAVD 88 (published height attribute is “Adjusted”) or was it determined by precise leveling performed by horizontal field party (published height attribute is “Leveling”).
The data sheet for station FA1337 states that the NAVD 88 attribute code is “GPS OBS.” [See box titled “Excerpt from NGS Data Sheet for PID FA1337.”] The data sheet for FA1560 states that the NAVD 88 attribute code is “Adjusted.” The orthometric height on the GPSBM file is different than the current published NAVD 88 orthometric height for station FA1337 (See table 3). This station’s leveling-derived orthometric height was superseded by a GNSS-derived orthometric height. Saying that, the GPSBM file only uses leveling-derived orthometric heights; therefore, stations that have been superseded by GNSS surveys are still included in the GPSBM file but their original published leveling-derived height is used for the analysis. Table 3 provides the orthometric height for FA1337 that was used in making GEOID12B. As previously mentioned, stations may be rejected by the geoid team based on the criteria outlined in the beginning of this column. Saying that, neither of the two stations were rejected by the NGS geoid team. This implies that the stations were consistent with their neighbors as far as the geoid model was concerned. Figure 6 confirms that all the stations around FA1337 and FA1560 are consistent with each other based on the Geoid16b_NAD geoid model. The fact that the two groups differ by 18 6 cm needs to be investigated.
Excerpt from NGS Data Sheet for PID FA1337
PROGRAM = datasheet95, VERSION = 8.9.1
1 National Geodetic Survey, Retrieval Date = OCTOBER 3, 2016
FA1337 ***********************************************************************
FA1337 HT_MOD – This is a Height Modernization Survey Station.
FA1337 DESIGNATION – RU 36
FA1337 PID – FA1337
FA1337 STATE/COUNTY- NC/RUTHERFORD
FA1337 COUNTRY – US
FA1337 USGS QUAD – FOREST CITY (1993)
FA1337
FA1337 *CURRENT SURVEY CONTROL
FA1337 ______________________________________________________________________
FA1337* NAD 83(2011) POSITION- 35 18 08.14237(N) 081 51 17.93516(W) ADJUSTED
FA1337* NAD 83(2011) ELLIP HT- 249.869 (meters) (06/27/12) ADJUSTED
FA1337* NAD 83(2011) EPOCH – 2010.00
FA1337* NAVD 88 ORTHO HEIGHT – 281.79 (meters) 924.5 (feet) GPS OBS
FA1337 ______________________________________________________________________
Figure 6 – GPS on BMS Residuals: Large Relative Differences Between a Group of Stations at the North Carolina/South Carolina Border (note: rejections by geoid team have been removed)
Figure 7 is a plot of the GPS on BMS residuals using Geoid16b_NAD that depicts a large difference between two stations only 20 km apart near the Maryland/West Virginia border. I will use this station pair to demonstrate the next step in my process.
Step 4 is to use the station’s NGS data sheet to determine the adjustment date the of station’s published NAVD 88 orthometric height.
The NAVD 88 attribute on the NGS data sheet states that both of these stations are coded as “Adjusted” but station JW0639 adjustment date is April 1995 (see box titled “excerpt from NGS Data Sheet for PID JW0639”) and JW1296 adjustment date was in June 1991 (the General Adjustment of NAVD 88). These large relative differences could be due to inconsistencies between adjusted heights due to the adjustment distribution corrections and/or constraints imposed in the April 1995 adjustment. Bench marks near the stations should be observed to determine if the same large relative difference exists, and the 1995 NAVD 88 adjustment project report should be reviewed to determine if a large distribution correction was applied.
Excerpt from NGS Data Sheet for PID JW0639
1 National Geodetic Survey, Retrieval Date = OCTOBER 3, 2016
JW0639 ***********************************************************************
JW0639 CBN – This is a Cooperative Base Network Control Station.
JW0639 DESIGNATION – J 17 RESET
JW0639 PID – JW0639
JW0639 STATE/COUNTY- MD/GARRETT
JW0639 COUNTRY – US
JW0639 USGS QUAD – ACCIDENT (1994)
JW0639
JW0639 *CURRENT SURVEY CONTROL
JW0639 ______________________________________________________________________
JW0639* NAD 83(2011) POSITION- 39 37 53.59739(N) 079 18 57.44776(W) ADJUSTED
JW0639* NAD 83(2011) ELLIP HT- 701.266 (meters) (06/27/12) ADJUSTED
JW0639* NAD 83(2011) EPOCH – 2010.00
JW0639* NAVD 88 ORTHO HEIGHT – 732.713 (meters) 2403.91 (feet) ADJUSTED
JW0639 ______________________________________________________________________
*
*
*
JW0639
JW0639.The orthometric height was determined by differential leveling and
JW0639.adjusted by the NATIONAL GEODETIC SURVEY
JW0639.in April 1995.
JW0639
Figure 7 – GPS on BMS Residuals Using Geoid16b_NAD: Large Relative Difference Between Stations About 20 km Apart Along MD/WV Border (note: rejections by geoid team have been removed).Figure 8 – GPS on BMS Residuals Using Geoid16b_NAD: Large relative Difference Between Stations 15 km Apart in Randolph County, West Virginia (note: rejections by geoid team have been removed).
Figure 8 is a plot of GPS on BMS residuals using Geoid16b_NAD that depicts a large relative difference between stations 15 km apart in Randolph County, West Virginia. This plot involves station HW3677 which has a published NAVD 88 attribute of “Leveling.” (See box titled “Excerpt from NGS Data Sheet for PID HW3677.”) The excerpt from the data sheet has the following statement: “The orthometric height was determined by differential leveling. The vertical network tie was performed by a horz. field party for horz. obs reductions. Reset procedures were used to establish the elevation.”
It would be useful if stations near this station were observed by GNSS surveys to determine what is occurring in this region.
Excerpt from NGS Data Sheet for PID HW3677
1 National Geodetic Survey, Retrieval Date = OCTOBER 2, 2016
HW3677 ***********************************************************************
HW3677 DESIGNATION – GPS 1
HW3677 PID – HW3677
HW3677 STATE/COUNTY- WV/RANDOLPH
HW3677 COUNTRY – US
HW3677 USGS QUAD – MILL CREEK (1995)
HW3677
HW3677 *CURRENT SURVEY CONTROL
HW3677 ______________________________________________________________________
HW3677* NAD 83(2011) POSITION- 38 37 50.21531(N) 079 55 29.64175(W) ADJUSTED
HW3677* NAD 83(2011) ELLIP HT- 1129.355 (meters) (06/27/12) ADJUSTED
HW3677* NAD 83(2011) EPOCH – 2010.00
HW3677* NAVD 88 ORTHO HEIGHT – 1159.91 (meters) 3805.5 (feet) LEVELING
HW3677 ______________________________________________________________________
*
*
*
*
HW3677 HW3677.The orthometric height was determined by differential leveling.
HW3677.The vertical network tie was performed by a horz. field party for horz.
HW3677.obs reductions. Reset procedures were used to establish the elevation.
HW3677
Figure 9 is a GPS on BMS residual plot of large relative stations about 30 km apart in Wasco County, Oregon. This plot has two stations with large differences and both stations have the NAVD 88 attribute of “Adjusted.” Their NGS data sheet states that they were both established in the general adjustment of NAVD 88 in June 1991. In this particular case, the leveling in this region is very old. As described in Part 7 (June 2016), you can retrieve all project identifiers for those projects with observations to or from a station using the station’s PID. The output from the NGS Data Sheet Mark Source Routine for PID RC1228 is shown in the box titled “Output from NGS Data Sheet Mark Source Routine.”
Output from NGS Data Sheet Mark Source Routine
Program: mark_sources Version: 3.0 Date: May 1, 2013RC1228OR/065 J 108
———————————————————-
GPS_OBS
———–
GPS_OBS FORE_POINT in GPS1655
DIR_OBS
———–
DIST_OBS
———–
VERT_OBS
———–
LEV_OBS
———–
LEVEL_OBS
———–
LEVEL_OBS STAND_POINT in L3410
LEVEL_OBS FORE_POINT in L3410***********************************************************
Figure 9 – GPS on BMS Residuals Using Geoid16b_NAD: Large relative stations about 30 km apart in Wasco County, Oregon (note: rejections by geoid team have been removed).
Figure 9 is a GPS on BMS residual plot of large relative stations about 30 km apart in Wasco County, Oregon. This plot has two stations with large differences and both stations have the NAVD 88 attribute of “Adjusted.” Their NGS data sheet states that they were both established in the general adjustment of NAVD 88 in June 1991. In this particular case, the leveling in this region is very old. As described in Part 7 (June 2016), you can retrieve all project identifiers for those projects with observations to or from a station using the station’s PID. The output from the NGS Data Sheet Mark Source Routine for PID RC1228 is shown in the box titled “Output from NGS Data Sheet Mark Source Routine.”
Excerpt from NGS Data Sheet for PID RC1228
PROGRAM = datasheet95, VERSION = 8.9.1
1 National Geodetic Survey, Retrieval Date = OCTOBER 2, 2016
RC1228 ***********************************************************************
RC1228 DESIGNATION – J 108
RC1228 PID – RC1228
RC1228 STATE/COUNTY- OR/WASCO
RC1228 COUNTRY – US
RC1228 USGS QUAD – WAPINITIA (1996)
RC1228
RC1228 *CURRENT SURVEY CONTROL
RC1228 ______________________________________________________________________
RC1228* NAD 83(2011) POSITION- 45 06 49.69715(N) 121 19 19.81396(W) ADJUSTED
RC1228* NAD 83(2011) ELLIP HT- 624.596 (meters) (06/27/12) ADJUSTED
RC1228* NAD 83(2011) EPOCH – 2010.00
RC1228* NAVD 88 ORTHO HEIGHT – 646.140 (meters) 2119.88 (feet) ADJUSTED
RC1228 ______________________________________________________________________
*
*
*
RC1228
RC1228 HISTORY – Date Condition Report By
RC1228 HISTORY – 1934 MONUMENTED CGS
RC1228 HISTORY – 1985 MARK NOT FOUND USPSQD
RC1228 HISTORY – 1985 MARK NOT FOUND USPSQD
RC1228 HISTORY – 20001010 GOOD OR-065
Figure 10 – GPS on BMS Residuals Using Geoid16b_NAD: Large relative Differences between Stations along the Oregon/Washington Border (note: rejections by geoid team have been removed).
Figure 10 is a plot of GPS on BMS residuals using Geoid16b_NAD depicting large relative differences between stations along the Oregon/Washington State border. It is the near Puget Island along the Columbia River. Station SC0330 and SC1086 are only 7 km apart and the relative difference is -20 cm (-11.4 cm minus 8.6 cm). This could be an issue with the NAVD 88 network design because there doesn’t appear to be many river crossing along the river between border stations. The fact that the residuals on the Washington State side are negative and the Oregon State side are positive is an indication that the stations need to be investigated.
Figure 11 – GPS on BMS Residuals Using Geoid16b_NAD: Large Negative Residuals North of Border between Oregon and Washington and Positive (or Small Negative) Residuals South of Border (note: rejections by geoid team have been removed).
The last figure, figure 11, is a plot of the GPS on BMS residuals using Geoid16b_NAD model that depicts large negative residuals north of the border between Oregon and Washington and positive (or small negative) residuals south of the border. This plot shows that the northern side of the river has large negative residuals all the way to the Pacific Coast. Once again, this is an indication that this portion of the NAVD 88 network should be investigated.
This column has focused on analyzing NGS’ GPS on BM data set that is used to make NGS’ hybrid geoid models. It provided procedures that users could employ when analyzing the differences between the modeled geoid values and the computed geoid values using GPS/Leveling data. This GPSBM data set or one similar will be used to make the next hybrid geoid model, as well as provide input to the transformation model between NAVD 88 and the new 2022 Vertical Reference System. All geospatial users should help develop this GPS on BMS data set to help improve the National Spatial Reference System and future hybrid geoid models. This column provided several examples of large relative differences in residuals between neighboring stations. Each example represents stations that should investigated based on different reasons, such as a weak NAVD 88 leveling network design in the region, the station’s published height attribute code implies that the station was not rigorously adjusted into the NAVD 88, and station pairs have different adjustment dates indicating a possible adjustment distribution correction issue or movement.
NGS has a program called “GPS on Bench Mark” to support users that occupy bench marks with GNSS equipment. This web site contains a lot of good information and provides the users with methods to recover, observe, and report information about stations in NGS’ database. The write up from the webpage is given below. I have highlighted a few sentences that the reader may find useful.
Improve the National Spatial Reference System (NSRS):
Recover: Look up the description of an existing bench mark and visit the bench mark of your choice. Observe: Record field notes, take digital photos, and collect GPS observations or coordinates for the bench mark you visit. Report: Use online tools to send the information to NGS.
Where?
Currently there are over 400,000 bench marks across the Conterminous United States (CONUS), Alaska, Hawaii and all U.S. territories. Tidal marks and bench marks are used for determining heights. Use the maps to prioritize which bench marks to observe.
Who can participate?
Anyone with Global Positioning System (GPS) enabled phones, hand held devices or survey-grade GPS receivers can participate. Recommended procedures vary depending on the type of equipment used.
When should I start?
You can collect and share information any time. Join volunteer efforts across the United States in celebration of National Surveyors Week beginning March 20, 2016. Contact the local National Society of Professional Surveyors chapter or your NGS geodetic advisor to learn about projects being planned in your local area.
By providing GPS on benchmarks today you can help NGS improve the next hybrid geoid model, increasing access to NAVD 88, and enabling conversions to the new vertical datum in 2022.
You can also help the local surveying community know about nearby marks by improving scaled horizontal positions and updating the mark condition or description by submitting a mark recovery.
What happens next?
NGS will use your data to update its databases and improve future models and tools. If you still have questions, contact the GPS on BM Team.
In addition to participating in the NGS’ GPS on Bench Mark program, all geospatial users should participate in NGS’ 2017 geospatial summit, which will be held in April in Silver Spring, Maryland.
This summit is an opportunity for all users of the National Spatial Reference System (NSRS) to obtain a better understanding of NGS’ plans to modernize the NSRS. Users will be able to provide feedback directly to NGS leadership. My next column will address NGS plans to replace the North American Vertical Datum of 1988 in 2022.
UAVs, precision agriculture and robotic guidance require high accuracy at low cost.
Emerging high-volume markets call for RTK technologies previously limited to niche markets by complexity and cost. This article discusses design and implementation of a very precise RTK-based module solution while maintaining cost, size and power consumption as low as possible. Several tests under a range of signal environments benchmark the new module’s performance against existing L1 RTK products.
Real-time kinematic (RTK) positioning has matured over the last few decades into a well-understood technology that, to date, has remained confined to high-end applications by high costs and complexity. Meanwhile, the rapid rise of robotic guidance applications has increased the need for higher accuracy for navigation purposes, fostering an ever-increasing demand for affordable and energy efficient high-precision solutions. Here we discuss the challenges associated with bringing RTK technology to mass markets.
The main challenge for any RTK receivers is resolving carrier-phase ambiguities to their integer values. To do so, an RTK receiver needs clean carrier-phase measurements. In general, high-end RTK receivers typically rely on multi-frequency, multi-constellation solutions and complex estimation models to improve ambiguity resolution performance. However, to reduce size, complexity and power consumption, mass-market receivers typically use narrowband single frequency front-ends, which increase noise and code multipath. Furthermore, mass-market GNSS modules have much less processor and memory resources to call upon. Therefore, to fully integrate the RTK engine, mass-market receivers typically need to restrict the computational burden by optimizing complex RTK algorithms.
Here we discuss our efforts to overcome these challenges while delivering centimeter-level positioning. Performance evaluation under challenging signal environments of a new mass-market L1 RTK module is benchmarked against an existing high-end L1 RTK product.
Multi-Constellation Support
A straightforward approach to improve reliability of the ambiguity resolution is to extend support to other constellations in addition to GPS. GLONASS and BeiDou have respectively reached full and initial (regional) operational status offering significant satellite availability improvements. Both systems broadcast their L1 open service signals using a frequency band that is offset with respect to that of the GPS L1 open service signals and, therefore, concurrent reception of GPS/GLONASS or GPS/BeiDou requires two distinct RF paths. Since the new L1-RTK based module can support reception of GNSS constellations using two independent RF paths, RTK support was implemented for both GLONASS and BeiDou, allowing either of these systems to be used with GPS. On the other hand, the low availability of operational Galileo satellites limits the benefits of a GPS/Galileo solution and, therefore, RTK support for Galileo was not implemented.
GLONASS Ambiguity Resolution
The Russian GNSS transmits L1 signals using a frequency division multiple access (FDMA) technique. While this increases the constellation’s resilience to narrowband interference, it creates two major problems for ambiguity resolution. First, GNSS pseudorange and carrier-phase measurements contain frequency dependent biases related to the receiver’s analog and digital hardware. For GPS (and other code division multiple access [CDMA]-based GNSS), all measurements share the same frequency and the biases cancel out during between-satellite differencing. However, this is not the case for GLONASS where the remaining inter-frequency biases are absorbed by the ambiguities, complicating their resolution. Second, GLONASS signal wavelengths are not common for all satellites within the L1 frequency band.
In addition to the double-difference ambiguity, GLONASS double-difference observations also consist of the between-receiver single-difference ambiguity related to the reference satellite scaled by the wavelength difference of the two signals.
Due to a lack of observability, the single-difference reference ambiguity cannot simply be estimated along with the double-difference ambiguity. On the other hand, merging the two ambiguity terms into a modified one results in an ambiguity that is no longer an integer and therefore cannot be fixed.
Both issues are well understood and several methods have been proposed to circumvent them. However, it is not yet clear whether the performance benefits brought by GLONASS ambiguity fixing outweigh the computational overhead.
BeiDou Ambiguity Resolution
China’s GNSS currently broadcasts B1 open service signals using mixed satellite and signal types, which could complicate ambiguity resolution. The limited orbit variability of BeiDou geostationary and inclined geostationary Earth orbit satellites produces poor carrier-phase ambiguity.
Despite this limitation, recent investigations reported very good dual- or triple-frequency GPS/BeiDou RTK performance, regardless of satellite type. Therefore our approach is to estimate BeiDou ambiguities for all satellites using appropriate weighting of the different carrier phase and pseudorange observations.
Cycle-Slip Detection
Single-frequency RTK inherently offers more limited measurement redundancy than its dual or even triple-frequency counterparts, making cycle-slip detection a difficult task. While a posteriori residuals checks provide a powerful mean to detect outliers, they are computationally expensive and therefore can only be used sparingly. To detect cycle slips prior to the measurement update, heuristic checks are performed on innovation sequences and complemented by systematic analysis of phase lock and C/N0 values.
Configuration Trade-Offs
The RTK positioning modules can concurrently receive and track up to two GNSS systems. By default, the reference receivers are configured for concurrent GPS and GLONASS reception. This can be modified to enable the combined use of GPS and BeiDou.
To optimize the use of processor and memory resources, the number of channels has been limited to 20. This is sufficient for dual-constellation operation almost everywhere except for a limited area in Asia where the number of visible GPS and BeiDou satellites can occasionally exceed 20.
Furthermore, the rover receiver can operate in RTK fixed or RTK float mode. In RTK fixed mode, the receiver will try to resolve ambiguities to their integer values whenever possible whereas in RTK float mode, the receiver will keep the ambiguity estimate as a floating number. The RTK fixed mode will provide the highest level of accuracy but can exhibit position jumps when transitioning from a float to a fixed solution or reliability issues when operating in degraded signal environments where multipath can lead to wrong ambiguity fixes. The RTK float mode, on the other hand, will typically provide dm-level accuracy but a much smoother trajectory.
Static Performance Evaluation
The static test data was collected on the roof of an office building in Singapore in April 2016. Twelve hours of data were collected by four receivers connected to a high-precision receiver forming zero-baseline for both GPS/GLONASS and GPS/BeiDou configurations. This allowed a thorough statistical evaluation of the ambiguity resolution performance for both configurations.
Static Data Processing
The static data sets were post-processed with a software using exactly the same algorithms as those embedded in the receivers’ firmware, allowing for direct comparison of different receiver configurations. The time-to-first ambiguity fix (TTFAF) is often used as a key indicator to assess the ambiguity resolution performance. The TTFAF differs from the time-to-first fix (TTFF) in that it only includes the time required by the ambiguity resolution algorithm to converge. To measure the TTFAF, the software is modified to perform a hot start (where position, time and ephemeris are kept) at regular intervals. This is done to increase the data set sample size and to provide a relevant statistical analysis of its reliability and rapidity.
Static Test Results
As expected, FIGURE 1 shows that the use of the GPS/BeiDou configuration significantly improves satellite visibility over the GPS/GLONASS configuration. The average number of navigation channels used is close to 20 when combining GPS and BeiDou whereas it remains below 16 when combining GPS with GLONASS. This produces faster TTFAF in GPS/BeiDou mode (FIGURE 2).
Walk Performance Evaluation
Two walk data sets were collected around Priory Park in Reigate, England on October 2015 and February 2016. Approximately one hour of data was collected each time with the equipment depicted in FIGURE 3. The antenna was mounted on a survey pole to ensure the best sky visibility possible. The radio frequency (RF) signal was then split three-way and distributed to a high-precision receiver, our rover receiver and a record and replay simulator. The RTCM correction stream was generated by a high-precision receiver connected to an antenna located on the roof of an office building and made available on a server. Using a Raspberry Pi and a 3G modem the RTCM stream was forwarded to both our receiver and the recorder. As shown in FIGURE 4, the Priory Park was selected because it provides excellent satellite visibility and is located approximately one kilometer away from the the reference station. While the open-sky test aimed at evaluating the performance of the RTK engine under ideal conditions, the tree-loop test was carried out to assess its ability to recover from moderate signal degradations. To this end, several loops were performed through the trees shown in FIGURE 5. [Click on an image to enlarge it.]
FIGURE 1. Number of satellites used vs. time for GPS/GLONASS (top) and GPS/BeiDou (bottom).
FIGURE 2. Zero-baseline TTFAF in Singapore.
FIGURE 3. Walk test set-up.
FIGURE 4. Open-sky walk test in Reigate.
FIGURE 5. Tree-loop walk set in Reigate.
Walk-Test Data Processing
The walk-test data sets were post-processed with a software using the same algorithms as those embedded in the receiver’s firmware. For the tree-loop walk test, the default GPS/GLONASS RTK fixed (Fxd-GR) configuration was used. The reference trajectory was obtained by post-processing the raw measurements from the high-precision rover and reference receivers with NovAtel GrafNav software. As it relies on a forward/backward post-processed dual-frequency GPS/GLONASS RTK solution, the reference trajectory is expected to be reliable and cm-level accurate. It can then be used to evaluate ambiguity resolution performance and baseline accuracy. Additionally, the recorded scenarios were replayed to a high-precision receiver. This receiver has an L1 RTK engine that supports GPS, GLONASS, BeiDou and Galileo constellations and is expected to deliver 1-2 cm positions. While this receiver addresses high-end markets, it was used to benchmark the performance of our RTK solution. Since the high-precision receiver supports the BeiDou and Galileo constellations using proprietary correction messages and not RTCM multi-signal messages (MSM), this direct comparison was only done for the GPS/GLONASS configuration using RTCM RTK messages. The high-precision default configuration will hereafter be referred to as Fxd-GR. The receiver was configured to output, amongst other, the NMEA global positioning system fix data (GGA) message which contains latitude, longitude and altitude data, as well as a quality indicator that can be used to see whether the receiver has achieved an RTK fixed solution.
Limitations of Walk-Test Setup
To generate a reliable and robust reference trajectory, a high-end dual-frequency wideband antenna was used. The antenna has excellent inherent multipath mitigation and phase center stability which is not representative of mass-market applications where the use of affordable patch antennas is likely to result in higher code multipath and lower C/N0. However, these issues can be efficiently mitigated by the use of a ground plane and a carefully selected reference antenna site.
Walk-Test Results
The open-sky walk test was performed in a location with clear satellite visibility so that the number of satellites with continuous phase is close to 20 during most of the test. Continuous phase lock is defined as the amount of time during which the receiver is able to track the satellite using a phase lock loop (PLL). Any interruption in PLL tracking is likely to trigger a reset of the ambiguity estimation. As can be seen in FIGURE 2, ambiguity resolution can take up to a minute, even for zero baselines. As such, having continuous tracking for longer time intervals is required to achieve high rates of RTK fixed solutions. As can be seen in FIGURE 6, this translates into cm-level position errors. Note that the open-sky walk in Reigate started and ended in an office area with low-rise buildings. The degradations brought by these buildings can also be clearly observed in FIGURE 6.
During the tree loop test, signal degradations caused by trees are experienced by the receiver approximately every five minutes, causing the number of satellites to drop to zero at regular intervals.
FIGURES 7 and 8 show the resulting position error for the mass-market and high-precision RTK receivers in Fxd-GR mode. The corresponding position error statistics are summarized in TABLE 1. The statistics are computed over the entire duration of the test and therefore can include position fixes that are computed using code differential or RTK float mode. While the large position errors that sometimes occur in these modes will tend to dominate the statistics, they are deemed representative of field applications.
Both receivers exhibit similar accuracy when they can fix ambiguities but the high-precision receiver sometimes recovers faster from signal loss-of-lock than the mass-market receiver.
UAV Performance Evaluation
A UAV data set of approximately half an hour was collected around a farm in Reigate, England in April 2016. The UAV test duration is effectively limited by the capacity of the UAV’s battery which, with the payload deployed for this test, was limited to less than 15 min. To extend the test duration, approximately 10 min of static data was recorded at the beginning of the flight while the UAV was standing in the middle of the field with no obstruction around it. The data collection was performed with DJI S900 hexacopter shown in FIGURE 9 and a payload similar to that depicted in FIGURE 3. The patch antenna was mounted on ground plane with a 15 cm diameter to mitigate multipath effects and ensure the best signal reception possible. The RF signal was then split two-way and distributed to our rover receiver and a record and replay simulator. The RTCM correction stream was generated by a high-precision receiver connected to an antenna located on the roof of an office building in Reigate and made available on a server. Using a Raspberry Pi and a 3G modem the RTCM stream was forwarded to both our receiver and the recorder. This farm provides clear satellite visibility and is located approximately three kilometers away from the reference station. It meets all the regulatory requirements to recreationally fly a UAV. The tree-line test was carried to assess the ability of our RTK engine to recover from moderate signal degradations and dynamics. To this end, the UAV was flown repeatedly along the tree line shown in FIGURE 10. [Click on an image to enlarge it.]
FIGURE 6. Position errors vs. time during the open-sky walk test in Reigate for mass-market receiver in Fxd-GR mode.
FIGURE 7. Position errors vs. time during the tree-loop walk test in Reigate for mass-market receiver in Fxd-GR mode.
FIGURE 8. Position errors vs. time during the tree loop walk test in Reigate for high-precision receiver in Fxd-GR mode.
FIGURE 9. UAV test set-up.
FIGURE 10. Tree-line UAV test in Reigate.
Test Data Processing
The UAV test data was processed in a similar fashion as the walk-test data. Two additional configurations, namely GPS/GLONASS RTK float (Flt-GR) and GPS RTK fixed (Fxd-G) were tested with the aim of illustrating their benefits and drawbacks. Due to payload weight restriction, it was not possible to embark a dual-frequency receiver for reference trajectory generation. Instead, the single-frequency raw measurements generated by the mass-market receiver were used. Recorded scenarios were replayed to a survey-grade receiver for performance benchmarking.
The main limitation of the UAV test setup is that the generation of the reference trajectory relies on raw measurements from our narrow-band single frequency rover receiver.. The lack of measurement redundancy and the increased probability of code multipath make the reference trajectory less reliable than that used during the walk test. However, UAV applications typically enjoy more favorable signal environment than their pedestrian counterparts. Additionally, it is possible to confirm the reliability of the reference trajectory using both the GrafNav backward/forward processing option and the reported accuracy.
However, the patch antenna used during the UAV test campaign is representative of mass-market applications. In fact, some tests have been conducted to compare the performance that could be achieved with various antenna types including, but not limited to, a high-precision antenna without its casing and a patch antenna with and without ground plane. The details of this investigation are beyond the scope of this article. Suffice to say that the performance of the patch antenna with a reasonably sized ground plane (15 cm in our case) was deemed the best compromise for mass-market applications in terms of size, weight and cost.
During the tree-line test, moderate signal degradations caused by trees are experienced by the receiver which cause the number of satellites to decrease at regular interval. [Click on an image to enlarge it.]
FIGURE 12. Position errors vs. time during the tree line UAV test in Reigate for high-precisions receiver in Fixed-GR mode.
FIGURE 14. Position errors vs. time during the tree line UAV test in Reigate for mass-market RTK receiver in Fixed-G mode.
FIGURE 13. Position errors vs. time during the tree line UAV test in Reigate for mass-market RTK receiver in float-GR mode.
FIGURE 11. Position errors vs. time during the tree-line UAV test in Reigate for mass-market RTK receiver in Fixed-GR mode.
FIGURES 11 to 14 show the resulting position error for the mass-market and high-precision receivers in Fxd-RD mode as well as those for the mass-marekt reeiver in Flt-GR and Fxd-G modes. The corresponding position error statistics are summarized in TABLE 2. Once again, this table can include position fixes that computed using code differential or RTK float mode.
Comparing the performance of the receivers in Fxd-GR mode, it can be seen that both receivers exhibit similar accuracy when they can fix ambiguities that the high-precision receiver suffers from an erroneous ambiguity fix at take-off which is also reflected in the position error 95 and 100 percentiles.
In Flt-GR mode the mass-market receiver is able to rapidly converge to dm-level accuracy. It is able to maintain this level of accuracy throughout the entire duration of test, highlighting the potential benefits of this mode for applications that do not require the highest level of accuracy but rely on smooth trajectory for guidance control.
For this test the mass-market receiver is able to fix ambiguity as often in Fixed-G mode than in Fixed-GR mode which is linked to the excellent satellite availability in the context of UAV applications. Additionally, the passes that were done close to the tree line were only performed later in the test, when ambiguities had already been fixed. This demonstrates the robustness of u-blox’s RTK engine to mild signal degradations. As a result, the NED position errors in Fxd-G mode are on par with those of the Fxd-GR mode. This highlights the potential benefits of this mode for high-dynamic applications that require higher navigation rate and operate in favorable signal environments.
[Click on an image to enlarge it.]
TABLE 2. NED position error statistics during the tree line UAV test in Reigate for different receivers and receiver configurations.
TABLE 1. Position error statistics during the tree-loop walk test in Reigate for different receivers.
Conclusion
Static tests showed that with fewer than 20 tracking channels, a single frequency GPS/GLONASS or GPS/BeiDou RTK receiver can successfully fix ambiguities in a reasonable time frame. During the walk and UAV tests, the performance of the mass-market receiver is similar to that of high-end receivers with respect to position accuracy and availability. For example, the availability of the RTK fixed solution was shown to be excellent under open-sky conditions for both but, as expected, in presence of moderate signal degradation and increased receiver dynamics, the availability of the RTK fixed solution decreases in a similar way for both receivers.
The kinematic data sets also served to demonstrate the versatility of the new mass-market receiver’s RTK solution. More specifically, the usefulness of the float-only solution for applications that do not require the highest level of accuracy but rely on smooth trajectory for precise guidance was shown. Similarly, the value of the GPS-only solution for high-dynamic applications operating in favorable environment was highlighted.
Finally, it is important to remember that while the walk-test results shown were obtained using high-end antennas, the UAV test results were obtained using a low-cost patch antenna, validating the suitability of RTK technology for affordable mass-market applications.
Acknowledgments
The authors thank Oscar Miles for his support with the data collection efforts in Reigate, and Alex Parkins for his contributions to the design and implementation of the RTK engine.
Manufacturers
The mass-market receiver described here is manufactured by u-blox. The RTK technology comprises a rover (NEO M8P-0) and a reference station (NEO M8P-2).
NVIDIA and TomTom announced they are partnering to develop artificial intelligence to create a cloud-to-car mapping system for self-driving cars.
The work combines TomTom’s HD map coverage, which spans more than 120,000 kilometers of highways and freeways, with the NVIDIA DRIVE PX 2 computing platform. Together, the solution accelerates support for real-time in-vehicle localization and mapping for driving on the highway.
“Self-driving cars require a highly accurate HD mapping system that can generate an always up-to-date HD map in the cloud,” says Rob Csongor, vice president and general manager of Automotive at NVIDIA. “DRIVE PX 2 for AutoCruise provides TomTom with a real-time, in-vehicle source for HD map updates.”
The NVIDIA DriveWorks software development kit now integrates support for TomTom’s HD mapping environment. The open solution is available for all automakers and tier 1 suppliers developing autonomous vehicles.
Low-cost, precision GNSS receivers will become a reality in the driverless car, drone and even smartphone markets by 2021, finds ABI Research. The automotive industry will be the main driver behind precision GNSS receiver adoption, in which centimeter-level accuracy is essential to complete driver safety systems with the redundancy necessary for autonomous vehicles.
“There is a variety of competing technologies currently under investigation by the automotive industry, but ABI Research forecasts it will move to a hybridized approach, combining LIDAR, HD maps, sensor fusion, machine vision and precision GNSS,” says Patrick Connolly, principal analyst. “As the receivers’ average selling price drops below $50, we expect to see a more immediate market for location technology services, such as AR Heads Up Displays (HUDs), in high-end vehicles. Vehicle-to-Vehicle, or V2V, communication might constitute another use case for high-precision GNSS.”
In addition to autonomous vehicles, the report also identifies opportunities for low-cost, precision GNSS receivers in autonomous unmanned vehicles (AUVs), as well as commercial and consumer devices. Though the average selling prices of such GNSS receivers is $1,000 and higher, ABI Research finds the cost to be one of the most addressable inhibitors to market growth today.
“Precision GNSS achieves sub-meter accuracy through a variety of methods, including a network of reference stations,” Connolly says. “The biggest question mark today is not cost-related, but instead how to achieve reliable, worldwide satellite navigation coverage to support correction techniques, such as real time kinematic, or RTK, and precise point positioning, or PPP. This is an extremely expensive undertaking, with currently no guarantee of a return on investment.”
Competition in the location technologies market ranges from crowdfunded startups to Internet giants, reflecting the scale of the opportunity. Traditional precision GNSS receiver vendors like NovAtel have the intellectual property, engineering experience and ownership of correction networks.
In the consumer GNSS receiver market, u-Blox and Skytraq lead the way, according to the report. Each developed low-cost single frequency PPP and RTK receivers, with a clear roadmap toward dual-frequency. Other consumer GNSS providers, like ST Microelectronics, Broadcom and Qualcomm, also appear active in this space.
Start-ups like North Surveying, NVS Technologies, REACH, and Swift Navigation continue to disrupt the industry, bringing low-cost precision receivers to market, said ABI Research. Their goal is to hit an ASP below $100 in the near future. And Radiosense is a startup that received a lot of attention for its previous work concerning precision GNSS on smartphones. It is now working on automotive solutions in a pilot in Austin, Texas.
Locata has the potential to be the wildcard in the deck, working on a powerful synchronization and location technology that may find its way into consumer technologies by 2021.
“Most interesting in the location technology competitive landscape is the involvement of Internet giants Google and Alibaba,” concludes Connolly. “Google recently announced it will make GPS pseudoranges available to developers, which, although extremely nascent, could open up the door for a lot of innovation. And in China, Alibaba is a major partner in the roll-out of Continuous Operating Reference Stations, or CORS, networks in the region.”
Transeeker is a pallet equipped with u-blox’s cellular and positioning technologies for accurate tracking: u-blox cellular UMTS/HSPA(+) module LISA-U200-62S and u-blox 7 standalone GNSS module, EVA-7M.
