The announcement of a new flagship rover for Trimble was met with positive chatter, but with a lot of questions as well. So, xyHt called the product manager…
In short, the familiar form factor of the R10 (slim bucket on a pole) is still there, but in the seven years since it was introduced it has taken two significant evolutionary steps: hardware, then software. A lot has changed in those seven years, especially with the near-completion of new constellations, bringing new possibilities for integrating many signals into GNSS solutions.
Announced on November 5th, 2019, the news generated a lot of chatter on surveying social media. Some was positive chatter, but also a lot of questions, and mainly those questions focused on: what are the major differences from the previous models, how does the new RTK engine work, and “Hey, I just bought an R10-2. What do I do now?”
R10 to R12 Evolution
There is a pretty large community of R10 users, it being a modern flagship rover that was launched in the middle of a growing boom in construction and development. A lot of people were buying a lot of gear, and many wanted to take advantage of the growing constellations. It could track (and include in solutions) satellites from four constellations (more if you count regional like QZSS); though Galileo and Beidou/Compass were at the time quite incomplete. We started seeing more rovers, like the R10, with hard alloy housings, and it also had tilt compensation, though only based on MEMS and a magnetometer for orientation (i.e. needed frequent user calibration).
Then, only a year ago, the R10-2 was announced. We interviewed the R10-2 product manager, Andrew Salmon then, and have again contacted him to answer the most common questions we are hearing about the R12.
“The R10 used a Maxwell 6 ASIC, “said Salmon “For the R10-2 we put in a new ASIC, the Maxwell 7, and we added more powerful processors, and RF modules where we made gains in power consumption.”
Understanding how a rover works is where such acronyms, and changes to hardware and software, would be helpful in decision making. Often such things are not presented to users who do not like to think their rover is just another black box. Now that RTK rovers have been around for over two decades, users have become quite savvy and want to know what is under the hood—and whether changes are significant enough to warrant upgrades. Salmon outlined how a modern rover works, and I’ve condensed it below.
The rover takes in the GNSS observations through its antenna as analog signals. The ASIC (application specific integrated circuit) does the analog to digital conversion and processes each signal from each satellite. That could add up to a lot of “channels” as you can have two to five signals from each satellite.
The processed observations, in the case of RTK and network RTK (RTN), are then combined in a filter base of an RTK engine (algorithmic, essentially a software component). As the mix of sats and signals gets richer, so does the premium on processing.
“The hardware upgrade in the R10-2 was done so with an eye towards the next generation of GNSS engine,” said Salmon. “The core positioning engine in the R12 is completely new.” It has been dubbed “ProPoint.”
So, the new R12 will sell for one price, but as it is essentially the same hardware as the R10-2 so that price difference between the two will be the same as a firmware upgrade for the R10-2. Same, except for the “R12” on the side of the housing. The firmware upgrade price depends on global market region; check with your local distributor, but I would not imagine it being more than a couple of thousand.
For a current R10-2 owner, “The upgrade is a simple option purchase, and you do a similar process as any firmware or option update in the receiver UI,” said Salmon. “The R10-2 will continue to get [unrelated] firmware upgrades, but ProPoint will have its own updates.” As the legacy R10 has the older hardware that can’t handle the new RTK engine, there is no direct upgrade path.
There are many approaches to integrating multiple constellations/signals into RTK solutions. Tracking is easy, processing in multiple channels is relatively easy, but putting them together in solution can require some heavy lifting. It looks good on paper to have piles of channels, but if not utilized in a solution then that is just a number. And you also need serious processing power to handle such mixes.
The R10 is capable of using multi-constellations/signals, provided the source of RTK or RTN corrections supports that. It’s like the standard RTCM3.2-MSM or Trimble’s own CMRx. Contrary to rumors, Trimble rovers are not limited to CMRx and can use RTCM3.2-MSM and other RTCM formats like most rovers. And Trimble VRS networks that output RTCM can support all rovers. But even though folks were able to use multi-constellation successfully with the R10, a method to optimize “big mixes” was already in the works.
“ProPoint was developed be constellation and signal agnostic,” said Salmon. “The engine does not care what input it gets, but obviously you need a minimum number overall.” You can work with any combination of GPS+GLN+GAL+BDS but could exclude any of them if desired or as needed. For instance, remember those incidents where the ephemeris for Glonass was messed up, rendering it useless for hours to days. Simply turn it off and run 1, 2, or 3 other constellations.
It seems odd to contemplate working without GPS, but in instances (though very rare) that someone is running one of those cheap “trucker jammers,” it usually only affects L1. There are lots of other signals in unaffected frequency bands, so you could turn that off and use the others.
You would need at least two sats of any constellation (to resolve its clock), but otherwise you can have combinations of say four from one constellation, from from another, three from another, etc. But, unless you’re in a very sky-challenged place, you will be using many more satellites, and I asked if for ProPoint “the more the better” is still worth it if you are in more open sky.
“Yes. In the past we used an approach called FAMCAR [Factorized Multi-Carrier Ambiguity Resolution) that was good and clever, that made use of limited processing resources,” said Salmon. “But now with the updated hardware, the idea of adding everything into one large filter base and having that choose the best for a solution is a reality.”
“When the user has more signals available in tough situations, even as signals attenuate, diffuse, or are obstructed, still many can be processed for an optimal solution,” said Salmon. “This is a different approach, and the [dev team] was also able to benefit from work they have done in advanced noise modeling.” Indeed, much more is known now about the nature of related things like differential code biases when working with multiple constellations, as has also benefited the implementation of multi-constellation solutions for RTN.
HD-GNSS is also implemented in the R12, as it had been for the R10 and R10-2. I’ve listened to explanations of HD-GNSS from some of the GNSS engineers, and while some of the deeper tech details make my head hurt, I like to use this shorthand: HD-GNSS is way to ensure quality over a whole range of solutions. It is not just tied to a Boolean “on/off” paradigm of fix/float where you may be limited if not “fixed.” Even if you are not looking for the highest precision, you could still have confidence in whatever range you are in.
I had to ask about tilt. Salmon said that the R12 has the same components as the R10 and R10-2 have with regards to tilt compensation: the MEMS and magnetometer. And he said that while one of Trimble’s construction rovers has some of the new generation of “no-calibration tilt,” surveying and construction can have different demands. He said there is no hard timeline and they are not going to announce until they are completely confident it would meet surveyor needs. Yes, we surveyors tend to argue about how many angels dance on the head of a PK nail. The R10, R10-2, and R12 have MEMS+magnetometer tilt, but do not have IMU’s at present. There are a lot of unknowns at this time as to the implementation of no-calibration tilt in the survey rovers and what upgrade and/or trade-in options might be if/when that happens. Again, this is something you would work out with your local distributor.
Salmon gave some advice on evaluating quality and good field practices to achieve it. This was more of general RTK advice, and not specific to the R12, but something to keep in mind especially when stepping into the world of many sats and signals. Most RTK users are aware of this, but it is good to reiterate.
“Some folks just fire up the rover and head off into the woods,” said Salmon. “But while you are moving, the engine may be loosened up to try to get a result. In Access you see the walking surveyor symbol and the standing one when initiating a measurement. It is when you start a measurement that you get the most realistic predicted results and quality indications.” This rang true; we often initiate a dummy shot just to see what we might get. “Think of it as different modes, like in a car that might have an economy mode and another for more power. When you start an observation, using multiple epochs is when you would be getting the best solution.”
Another question users often have about multi-constellation is, “How do I know how many sats are actually being used in the solution?” It is easy to see how many the rover can “see” or track. But let’s say you connect to a base or RTN—that number of sats in the measurement window in Access will drop to how many it has in common with what the base/RTN can see and resolve. Then when you initiate a measurement, the engine might evaluate and drop others. It is all becoming less of an issue with 4+ constellations. When folks are starting to see 30, 35, 40 in view, but even if it drops to 20 then it is still primo.
If you are going to test drive one (or for that matter any rover that can utilize multi-constellations), make sure you have a source of RTK corrections available, like a base, R12 or other, that supports and transmits RTCM3.2-MSM (Multi-Signal Message) or CMRx, or an RTN that provides the same.
To be fair, nearly every GNSS survey rover developer is working on doing multi-constellation/signal RTK integration, even some of the 4th wave GNSS off-brands that have varied success, from coarse, to “whoa!” Then there are others who have not chosen to work on this; either it’s not a priority, or they haven’t figured it out yet. The days of too-few Galileo and Beidou sats are over (as for instance Beidou launched as many as 12 sats in a single year.) Gone as well are the days when conventional wisdom was that GAL and BDS made no difference… I still hear that, and one think I heard was, “If someone hasn’t figured out how to make hammers, they might want you to question the value of nails.”
All of that aside, we do not endorse any products and do not do Consumer Reports-style, rigidly controlled tests—you will be the best judge when you test drive this or that.
I would like to make a note about the team that was chiefly responsible for the core of the new engine, the Terrasat team (Trimble Munich). Lead by Dr, Herbert landau, there are about 30 PhDs in that facility. Landau lead the development of the first commercially successful RTK system in the mid ’90s. The team developed VRS (now implemented in their Pivot RTN suite), HD GNSS, monitoring engines (e.g. T4D), and much of RTX (L-Band real-time and post-processed PPP). I have visited that team in the past, and apart from being a team of extremely bight developers, there is a deep desire for quality and keeping up the long and distinguished track record—they are not prone to hyping or overselling.
We’ll do our own eventual test drive (in the near future), but a network of advance users has had some time to try it out, folks who had been running R10s and/or R10-2s. A surveyor I know in South Africa was quite impressed. Lee Landman, owner/operator of LB Landman of Cape Town, said: “It definitely gets shots that I could not get before with an R10 in difficult environments. For example, the other day I tried to get three really difficult tree shots, the R12 got two of them outright and a compromise on the third tree, the R10 got none. So, it was two of three vs zero of three.” He also said he is about to try it out in more urban environments and is already planning to replace older rovers in the future.
You can find YouTube videos of some of these advance users and some of Trimble’s own field demos. Trimble says they see a 30% gain in tough environment performance over their own R10. You will be the best judge of the performance. Until you can test drive it yourself, the other big takeaway from this announcement is that it looks like GNSS development is definitely not stagnant, and we can expect much more.