I was wondering when someone was going to do this: No base, no RTN, no cell, no bubble? No worries. Then throw in photogrammetry on the rod.
There were many noteworthy surveying technology announcements in the past year; several in particular spurred me into the mode of, “Okay, I gotta try that.” Trimble announced two major enhancements to their R10 GNSS rover: a full-blown global, real-time PPP solution (Trimble CenterPoint™ RTX™—yes they did trademark those terms) and tilt compensation to the electronic bubble.
Then there is the V10, a pole-mountable 360º camera array for terrestrial photogrammetry (the first designed specifically for these types of workflows), which held the most surprises when I finally got my hands on one. I quickly found out that the real coolness is in the back end—the workflow. So, instead of a straight-up “box” review, I’ll focus on a workflow first and work backwards to the hardware.
V10: Photogrammetry on the Rod
Say you have a project site that might have one or more of the following challenges: tree cover, little control, mixed terrain, heavy traffic, features behind fences, and a limited window for performing the survey. To overcome the challenges, you seek to make as many useable observations as possible while meeting the accuracy required.
Multiple setups with robotic total stations traditionally fit the bill. While a GNSS rover alone might work in fairly open areas, it is not going to help much with the areas of the site under canopy (though technology is improving in this arena), and you may have to have to fill in with the total station, anyhow. You could try to shorten the time on site and reach into a sky-view-challenged or inaccessible area with a full-blown scanner.
Or you could try some DIY (do-it-yourself) terrestrial photogrammetry; there are a lot of cobbled-together, strap-a-camera-to-the-rod solutions out there. Some terrestrial photogrammetry solutions use off-the-shelf cameras and various free or cheap software packages, etc., with less-than-optimal results. But, to properly do terrestrial photogrammetry with any amount of confidence traditionally requires a lot of pre-planning for geometry, control, target points, and manual photo registration.
What if you could take an instant 360º panoramic photo from the point of view of any GNSS rover shot on the site (or potentially every shot)? You could choose from and process any of these shots that have overlapping photos of your areas of interest, then process and align these like an adjusted network.
There are several adjustment strategies to choose from: unadjusted raw locations; hold the internal compass azimuth; using photo observable backsights; or doing standard resections, which means you have control over which elements to hold, float, and adjust. This can be tough (or ill-advised) to do with DIY or low-end tools.
The Vision module for TBC (Trimble Business Center software) has menu-driven workflows for all of the above: raw only, azimuth, standard resection and full orientation, and if done with the optimal number of common tie-points (12 or more for standard resection), the automated tie-point recognition and adjustment can fully correlate and register the 360-pano images.
You could do photogrammetry with a minimum of two photo stations, but why skimp? The V10 takes the pano images in less time than the RTK shot, so fire away. You can store over 3,000 panos in the control unit. Even if you had previously set out points with a total station, you can follow up with the V10 by itself and take the pics.
There are 12 calibrated cameras on the V10: seven overlapping cameras arrayed to take the 360 and five below to include the view to the ground (in a direction and field of view you choose). The 5-megapixel images from each of the calibrated cameras are projected together, via the graphic card, into seamless 60-megapixel panos.
Unlike a lot of common “pano tools,” the images from the V10 are not “stitched together.” The cameras are calibrated and do not need any tweaking to fit. All 12 pano images are projected using the graphic card and the calibration data. The exporting function of TBC then puts them together. The automatic tie-point selection chooses visible points (those with sharply defined pattern/edges/corners/contrast and optimal geometry) common to multiple overlapping panos, adjusting per your desired method. The tie points are labelled, and you can add more manually and add observed tie points.
The measuring of a photo point in TBC, which the menu calls “MeasurePhotoPoint” (go figure), always uses the original image and the calibration data. From these calibrated and registered images you can use the MeasurePhotoPoint pixel selection tool in two or more panos to observe points for your topo, reducing time on the site, out of traffic, and from a laptop in the comfort of your truck or office.
Using this tool you can generate real-world X, Y, Z positions from the pictures. If you can see it in the picture, you can measure its position. Standard Feature libraries/codes can also be used to generate points, linework, and polygons. You also have the entire site documented (in case somebody missed something; like that has never happened!). The office folks can see everything the field crew could view.
The scenario I wanted to test was the topo of an intersection. One crew went out one (sunny and clear) day with a R10 GNSS rover mounted above the V10, which is a common configuration. (You can also mount a prism above the V10 for use with a robotic or conventional total station. If an object then is out of line of sight of the total station you can still capture it with the V10. They also took RTK shots on known points, several of which would also make good tie points and check shots, like the corners of stop-bars.)
I went out the next day (Denver’s first bitterly snowy day of the season) and repeated some of the work, checking into points they observed directly and some picked from the adjusted panos.
Taking panos with the V10/R10 configuration was as simple as a check box in the Trimble Access software on the tablet, the Trimble Yuma 2 tablet computer. The V10 only works with a tablet, but I appreciated the extra real estate when working with the images.
It took a pano when I took an RTK shot (I did learn to duck so as to avoid having my bald spot in each pano). I was able to preview the pano on the controller and zoom in to the individual images from each of the 12 cameras to see if it clearly shot what I was looking for—there are options for discard and retake.
This was all done without having to stand out in traffic. In this case, from just 12 photo stations (point with pano), the entire intersection could be surveyed.
The vertical was surprisingly better than the horizontal (for example, inverses consistently under 2cm horizontal and a little over 1cm in vertical). I asked Chad McFadden, a surveyor and portfolio manager for Trimble’s optical products, why the vertical is often better. “The concept makes sense after you see how the tie-point adjustment works,” he said. “The registration relies a lot on the vertical points.” This was for an intersection that was around 20m across.
The rule of thumb is “one in ten,” meaning you can expect precision of around 1cm per 10m away. If that sounds limiting, remember that you can take a lot of pics very rapidly. If features are too far away for the precision you want, walk forward and add more photo stations to the network. When in doubt, shoot more.
We also had a conversation about best practices for making sure there are good common tie points. Standard “good triangle” principles are used, and you can even put adhesive targets on the pelican cases and set them out at optimal distance/angles if there are few common features. Indeed, if on a flat-open expanse (like the area around that Denver intersection) where the only well-defined common observable tie-points may be too far away, they recommend sticking a bunch of lath in the ground around the site.
There are a lot of quality indicators to gauge the precision of your photo-selected points: you can see the residuals of adjustment, the strength of figures, and the estimated precisions throughout the processing. Again, most uncertainty can be addressed by taking more shots. This is a very compact and portable solution; the V10 only weighs about as much as an R10, and it has a recessed side so the UHF antenna of an R10 mounted above can point down, which is what we did on this project.
This brought me to the other two features to try out: tilt compensation and Trimble RTX. Read about them in the extended version of this article online.
RTX: Global Precise Point Positioning Service
Since its release about a year ago, R10 users have been able to get a taste of real-time RTX technology with the xFill feature. With xFill, messages from Trimble’s global network of tracking stations, which are modeled to deliver messages via communications satellites, can keep a real-time solution going for several minutes after a rover has lost its connection to a corrections source (e.g., cell drops). But now, full real-time CenterPoint RTX, like that enabled on the R10 (and available to all R10s via a firmware upgrade) takes this much further.
By using messages like enhanced clock and orbits, plus some other modeled data for things like ionosphere, the R10 can do Precise Point Positioning (PPP) from scratch, converging over 20-30 minutes, then enabling subsequent observations of around the 4cm precision range (sometimes better, but higher in vertical, like any other GNSS method).
This sounds like a long time, but think again about the workflow. If you do not have RTK/RTN available, you would have to observe, post-process, and then start taking observations by some other means. In contrast, you can do the RTX convergence in the time it typically takes to set up the gear at the site. If you start from a known point (they call it “quick start”), the convergence can be reduced to around five minutes.
While this does not sound precise (compared with RTK/RTN), keep in mind that this can be done with no base, no network, no cell—basically anywhere that you can see one of the satellites transmitting the Trimble RTX signals. Before someone says it: no, of course there are things you would never do with this, but I can think of great many things you could do with it.
To test RTX, I asked Richard Brush, surveyor and application engineer with Trimble, (their internal testing expert), to give me a crash course in the RTX service. It took just a few minutes. Considering the conditions were not great—open sky, but a near-snowstorm—the first convergence was under 18 minutes and came in at 4cm (h) x 7cm (v). Brush explained that progress bar for convergence would run until it reached a minimum of 7cm (h) x 12cm (v). Letting it run beyond the initial convergence saw the numbers improve; the best practice is “if you have the time, let it run.”
After convergence I could walk around the field and take more topo shots. Unless it lost view of the geo-stationary comms satellite for more than about 30 seconds, the position of those satellites shows up in the sky-view on the controller as well. Keeping connected to the geostationary sats down by the equator requires forethought and situational awareness, the major challenge in using this solution (or any real-time PPP solution that uses geostationary sats).
The quick starts took only about two minutes to converge. In a quick start a known position—from a previous convergence, RTK/RTN, or published mark—gives the RTX convergence a bit of a jumpstart on a convergence, and it will proceed to give a new position. The only caveat about the known position is that, at this time, it has to be in ITRF 2008 Epoch 2005. You can easily transform other values to ITRF 2008, and Brush explained that implementing on-board transformations is a priority.
There is a post-processed RTX service as well, and a lot of folks discovered this free service (real-time RTX is via subscription) for the first time when NGS’s OPUS was temporarily unavailable during the recent federal shutdown. In my own tests, and hearing from others, the results of this global RTX solution were in most cases comparable to OPUS (for the same length observations) but with one notable advantage over OPUS: multi-constellation. This includes not only GLONASS, but the QZSS system as well (of course there are only a few places in the continental US that can see QZSS, as it is doing that HEO; highly elliptical orbit, over Japan and Australia).
There is also a preview version of the post-processed service that uses as many Beidou and Galileo satellites as are available. Multi-constellation is a biggy in my neck of the (heavily forested, Pacific Northwest) woods; plus, in our coastal regions there are no CORS offshore for optimal OPUS geometry, so a global solution does have appeal.
Bubble and Tilt Compensation
The R10 was designed with an electronic bubble (yes, you can still have one on the pole). An electronic bubble can be precisely calibrated and configured to prohibit shots when out of tolerance or even to take shots automatically when within a chosen tolerance.
There is a lot of hand wringing about electronic bubbles, but compensators and electronic bubbles have been around in various forms for some time, and the R10 bubble has proven to be very useful and reliable. There is even a handy dimple on the top of the unit so you can test it with a plumb bob to your heart’s content (and also have a standard bubble on the rod to check against). It can save a lot of time on certain types of work.
Early in 2013 I took an R10 out on a topo of gravel road running seven miles around a reservoir. I walked in one direction with the R10, which was set to shoot when within vertical tolerance, and another surveyor took another new rover (without an electronic bubble) and walked in the other direction. Twice as much road was topo’d with the R10 in the same amount of time (we did swap to check). But we did wish for tilt compensation, not released at the time, for other types of work.
The new tilt compensation feature is available for R10s via a firmware upgrade. My tests (on the same snowy intersection) were very simple: shoot knowns, then reshoot while tilting the rod at various angles. The interface for the electronic bubble—its calibration and settings—are in the Access software on the controller. When you tilt the rod you see the position in a circle representing the tolerance you have set, and you see the position crossing concentric circles representing five-degree increments of tilt. You can only compensate for 15 degrees of tilt. In the tests, the inverse between the known and the tilt shots were under 1cm (h) and a few mm in vertical at 10 degrees; this went up to nearly 2cm (h) but still only a few mm in vertical for 15 degrees.
One question several surveyors wanted me to ask is, “How does the tilt compensation feature know the direction?” There is an electronic compass that you calibrate by rolling the R10 along a horizontal axis at eight points of the compass via preset routine, and there are multi-axis tilt sensors and accelerometers. While a rover can determine geodetic north via GNSS, especially if moving, the tilt is relative, and over such a short axis the direction of the inverse checks out quite well over longer baseline checks shots.
What can you do with tilt compensation? Get around the many annoying obstacles on a site. Shoot up against a fence, pole, or shrub. I really like the idea of being able to shoot inverses in catch basins and maintenance holes with less pain. Like the other features tested, I found more things to do with them once I got my hands on one.
“If you give people tools, [and they use] their natural ability and their curiosity, they will develop things in ways that will surprise you very much beyond what you might have expected.” —Bill Gates
Surveyors sometimes have a lot of time in the field to daydream about things that might make their jobs easier, but also things that might enhance what they are doing, broaden marketability, etc. I’ve done a lot of that daydreaming myself and have been listening to surveyors’ blue-sky ideas for cool tools and solutions. I’ve heard ideas like “flying total stations” … now we have UAS.
Several years ago, for a conference presentation about future surveying, I sketched an amalgam of many of these ideas (pardon the rough sketch). It was a bit tongue-in-cheek though, as I piled too many ideas on one pole. Some anticipated futuristic ideas include a scanning/imaging head, tilt compensation, voice commands, and heads-up display glasses (Google Glass anyone?). Several of these have come to fruition (sort of, but we still get to do a bit of a “told-you-so” dance).
There are a few other ideas that are a bit “out there.” These include gyro-stabilization, batteries and solar mesh on the vest (with induction gloves to keep the rover charged), a wand to direct scanning/imagery, a variable mask GNSS antenna, a phased array globe-antenna for anti-jamming-spoofing-interference-alien-rays (hee hee), and a feature so this antenna could work with terrestrial signal ranging (e.g. Locata, terralites).
My favorite (that gets the most flak): the gravity meter in the pole. Dream big. But now I’m finding out that hardware dreams may be only part of this enhanced future; it is not all about the “box.”
Like many other surveyors, I find myself focusing too much on the tool and not seeing past the physical box to the full potential it might have. There are lots of boxes and gizmos, and you can have the most technologically supercharged box in the world, but unless it can fit into an enhanced workflow, while respecting established standards and practices (for applicable surveying tasks), it is just a cool box. Most of the practical magic is actually in the backend. I get burned out on solutions looking for problems or niche tools.
Despite the initial cynicism we might have at some new tool, we are seeing a lot of truly practical gear from many of the manufacturer. It is encouraging to see emphasis on workflow enhancements rather than just “magic box” development (though some have not quite gotten that message). The profession is learning that the true value of a tool is how well it can fit into a workflow and mesh with other gear and office tools, but also how it can open new markets for surveyors’ services and solve problems we might not have thought possible.
Gavin Schrock, PLS is a surveyor, technology writer, and operator of an RTN. He’s also associate editor of this magazine.