Which program did you use for least square adjustment? Starnet?
SALSA, we have a customer that uses Starnet, we wrote software to convert target observations from Trimble jobxml and Topcon scan project files to SALSA and Starnet formats.
- I will do the laser scanning with a Leica P40.
- Its the first time that I am trying to establish a so precise 3D network.
- we have an upcoming project for scanning bridges combine it with photogrammetry. Also we have to scan a big industrial space all the above has horizontal and vertical accuracy 2mm. That’s drive me crazy but is challenge for me also.
Customer expectations can be painful. I suspect other companies are telling the customer they can achieve 2mm, or have told them that. Leica's P40 is one of the most accurate terrestrial scanners available on the market. But we're talking working in real world conditions, not a calibration laboratory. I would recommend adding wording to the effect of "as accurate as practically achievable." Achieving 2mm on your control network is difficult, let alone scan accuracy if I'm interpreting that right. The 3D positioning accuracy of the P40 is 3mm, its angular accuracy is 8" and range accuracy is 1.2mm + 10ppm (1 std dev) and a very good dual axis compensator for a scanner at 1.5" Metrology grade scanners may have micron range accuracy, but their angular accuracy is usually above 20 arcseconds, so they're only good at short range and are designed to operate in an enclosed controlled environment.
Positioning accuracy is key here, the 3mm 3D positioning accuracy is based on using scan targets to determine position, however the P40's position can be measured directly using a prism traverse kit and scan targets used for angular alignment only. I'd suggest large scan targets placed at distance beyond the work, so any angular errors are reduced.
You'll need to register using Cyclone Core, I doubt Reg360 has the capability.
These are some results from using a Topcon GLS, with scan positions determined by least squares. Scans are captured from inside and outside a large industrial shed with corrugated wall sheeting, you can see where the sheeting overlaps, there is no scan overlap, there is no direct line of sight between the positions either. The wall sheeting is approx 0.8mm thick.
The GLS is somewhat equivalent to the P40 in accuracy.
I have surveyor background as I do this for the 20 years now.
I'm a mechanical engineer with a background in field metrology, my daughter is a surveyor, we operate a small family business. 😉
@pfirmst Did you really have to roll your own software? - Trimble does have a stylesheet for export from Access to Starnet
smr prisms is a good solution already have one but I want to order two more.
SMR's are effectively our scan targets, so we have 20 of them, with all sorts of different mountings, from traverse kits, small magnet mounts, large magnet switch mounts (for pipe or tube), and mini ground tripods.
The advantage of using prisms instead of scan targets is better positioning accuracy, they can be shot from long distances, and low angles.
We set them up all around the job site and do rounds from multiple positions with an SX12.
We also use 62mm Sokkia AP01 prisms for long range > 150m, but have found the GLS scanner has vertical errors with them at close range.
Our gear is set up for air freight, for remote work.
@pfirmst Did you really have to roll your own software? - Trimble does have a stylesheet for export from Access to Starnet
Yes, for export to SALSA, also there's nothing on the market to convert Topcon GLS scan target observations. We're also including the scanner target obs in LSA.
We use a reinforced Topcon prism holder with a stainless-steel tube set in Devcon putty for the smaller antennas. These are older GPS only Leica units (serviced by Chris in Poland), we don't insure them against theft, but they do have i-tags, so far, we haven't had any go missing, touch wood. Beyond 24 hours duration, AUSPOS only accepts full days of data, we wrote software to clean up the RINEX files (remove incomplete records). https://www.ga.gov.au/scientific-topics/positioning-navigation/geodesy/auspos
Anubis is very useful for checking GPS observations: https://gnutsoftware.com/software/anubis/
AUSPOS will also reject submissions with too much multipath, which is why we're using ground planes and a choke ring antenna.
Long observation static GPS really helps remove any local distortions we would have if we fixed our LSA to local control, as the accuracy and precision desired by the customer was beyond the typical state plane network benchmark coordinate accuracy.
@pfirmst Do you send your SMR’s off along with any other prisms to be checked and calibrated. When I was doing metrology work we were required to have all SMR’s calibrated twice a year. What we found out is about 2 to 3 years deepen how well some took care of them they would fail. Most of the ones that failed you could see some knucklehead had dropped one. But sometimes with naked eye no visual could be detected so we would have to use those only for rough work. Love the SMR’s though. So many easy ways to mount and measure things with all the edge plates etc. it took my survey brain a while to stop thinking of every location as just a point though and start visualizing planes etc.
You have one of the most accurate geodetic laser scanners on the market, but there isn't any geodetic laser scanning hardware that can achieve the level of accuracy your customer wants, at least not outdoors, in real world conditions. It's not an easy conversation to have with your customer.
There are other things to consider too, such as spot size, emissivity, reflectivity and roughness of the work surfaces and angle of incidence of the laser. These all affect accuracy. Initially I couldn't find anything about the spot size of the P40, but now I've found it, it's 3.5mm at the front window, with a beam divergence of 0.23 mrad, the MS60 is 8x20mm @ 50m, the SX12 is 14mm dia @ 100m, the Topcon GLS is 11.3mm @ 150m, or 4.1mm in detail mode (which reduces range to below 100m).
The laser spot hits an area and the instrument records the average return from that area, if it's not flat, details below spot size can't be captured. Metrology scanners have sub mm spot sizes and very limited range as a result, but these might be used to target small critical areas, but aren't practical for large areas.
We also allow the instruments to warm up, measure pressure and temperature and enter it into our scanners and total station, as they're calibrated under controlled conditions of 20 degrees C and 1013 hPa, we often operate in environments between 30 and 40 degrees C and at lower pressures in higher altitudes. As far as I'm aware the Topcon GLS is the only scanner that uses temperature and pressure compensation, apart from scanning total stations. The further away the scanner is from calibration conditions, the more accuracy will be impacted.
Prisms provide much more reliable distance measurements.
I'd ask which areas specifically need to be 2mm and measure those with a total station & prism from at least two locations for LSA, they might be bolt locations and the distances between them might be critical relative to each other, rather than relative to state plane coordinates. Point clouds produced by geodetic laser scanners aren't much good for picking accurate bolt centres as they don't capture small enough detail, unless you're dealing with 2 in dia bolts.
Using LSA, we can get our occupations very tight, and we can select a prism observation with the lowest std-dev for our backsight in registration software and even select a scanner with the best calibration certificate.
Taking every care and precaution, I think we're looking at a range of 6 - 10mm 3D position accuracy @ 95% confidence for laser scanner points in registered point clouds (after trimming scan radius and removing outliers with filters on the registered point cloud), flat surfaces perpendicular to and in close range of the scanner with good emissivity will have the best results. Low angles of incidence aren't accurate, eg if you strike the ground at 10 degrees, you're getting the tops of rocks and gravel, but at 60 degrees you get right down between the gravel as well as on top, so there will be elevation errors, I could write a software filter to remove points with low angles of incidence, but it couldn't distinguish between rough and smooth surfaces and often we want to keep points from smooth surfaces at low angles, eg concrete or steel.
The vertical accuracy I've been quoting is ellipsoidal height, once it's converted to orthometric height, then we're talking cm or inches of vertical uncertainty, unless there's a 1st order fixed height datum within reasonable distance...
We had this discussion with our customer who had their own surveyors, who backed us, the customer had engaged an engineering company to design new plant and the unachievably high accuracy scanning standard was theirs. We still won the job, under the condition that what we produced was as accurate as practically achievable. It turned out to be a fun, really challenging job. There were times when were even hosing and cleaning the work are down, as areas of interest were often covered in product and we also had to recover the original plant coordinate system, which had us clearing lines of sight with an electric pole chainsaw through the scrub.
Thanks for the tip, we haven't been using them that long yet. 😉 We had been using 62mm dia Sokkia prisms with the GLS, but were experiencing vertical angle errors at close range, we had a job working for a Surveying company that supplied 1" dia SMR's and noticed we were getting good vertical angles in LSA, the GLS couldn't auto detect the 1" SMR's and it had to be manually pointed at the prism, so we tried 1.5" dia SMR's and the GLS was able to autodetect them. We've only ever used prisms with the GLS, but the scanners also came with scanner targets with round silver circles, the ones for close work were only 30mm diameter, so I suspect the 62mm dia was just to large for the scanner at close range, it's fine from 150 to 400m range, but below 135m range we start to see vertical errors. We haven't had the opportunity the use the 1.5" SMR's yet, but will in the near future.
We don't see the vertical errors with the total station on the Sokkia Prisms, so it's definitely something to do with the way the scanner is measuring them.
@pfirmst yeah some great tools for sure. Look up a software called Spatial anylizer . Great software and can be used with many total stations as well as laser trackers etc. with tight metrology work it can make for a much easier system than land surveying software. I believe they were working on the old sx10 being able to work with it but it has been some years about 5 now since I messed with metrology work. We had laser scanners terrestrial scanner Faro and sx10 at the time was in the que for purchasing along with some other surveying equipment and software. When I attended my class at the spark anylizer school which was about two weeks long I was amazed at how many different industries were utilizing it to make things or keep things in adjustment so manufacturing could be used. A high end tire company had prop in the class and the tolerances in which they were seeking to achieve amazed me. I mean a lot goes into making a tire that I would have never imagined. All these years doing surveying in one form or another I was humbled to the fact that there was this whole other world of precise measuring gurus out there.
you are correct in what’s achievable in reality outside vs inside. Working inside ships to on deck makes a huge difference. I did some work out doors certifying a certain item that was tough as sunlight and temp changes made for a long day at times. If indoors it could have been done was in a couple of hours. I worked hand and hand with mechanical engineers as items expanding based on surface temperature based on composite and my ability to have to change on the fly was tested daily. I could get something about right and then new computations would come out and bam I had to start all over. I learned a great deal from them. Hopefully I was able to give a little knowledge to them as well.
@olemanriver Thanks again, that's a great tip, just proves the value of sharing information on these forums. It's hard to find software that works with multiple vendors and I'm always keeping my eye out for good software. Vendor software usually has issues with other equipment. When we worked outside, we did so on the graveyard shift, we had to allow temperatures to equalize during the night, we'd take our final measurements around 4am just prior to sun-up, then after dawn everything went haywire and further measurements had to wait until the following night.
Something I had wanted to change in engineering & manufacturing was adding temperature ranges tolerance tables on machining drawings, especially when large components are made all around the world in different environment and are expected to fit together.
@pfirmst now on that spatial analyzer software it’s not a surveying software but for metrology. So a radial arm scanner laser trackers scanners certain types total stations etc. but since you are in both worlds it most definitely might be a good tool for you to have in the tool box. If it does have the drivers for your sx12 then that would be a great asset as you could turn compensating off in sx12 and basis it as if it was a laser trackers so that software also performs lsa for that realm. I also used meshy and some others to build meshes and bring the built mesh into spatial analyzer and get all the measurements qa/qc and any reverse engineering task the PE asked for. Then flap of that was taken into solid works and a in house program to build final products. Good luck as you build your company for sure. Always glad to help. I have been impressed with this site myself great knowledge base here. Great place to learn for sure.
@olemanriver We have Geomagic Design X and Solidworks licenses, but good to know we might be able to bring SX12 measurements into the metrology world. Useful if we need to have something on control, but also need relatively high local accuracy.
@pfirmst yeah I remember playing with geomatics design a little. In a very simplified way the sx12 running Trimble access can orient via a plane and so it has started its way into some of the metrology workflows. So if it could be run via solids spatial analyzer then it could be a perfect tool for tying in and bridging the gap of the survey control and your higher end precision stations for the metrology side.
I did have to design a transformation as I was on a datum but the mechanical engineers didn’t really are about lat longs nor state plane so for them I placed 0,0,0 at center of site and flowed out from there. As you know north east elevation is not necessary in a true 3d orthogonal system neither was orientation really it was all on its self. Another they survey brain had to deal with a was not always up. LOL. And they had the same problem thinking about gravity sometimes when there design worked but now had to be placed on the ground the younger folks didn’t understand I needed to know which way was up so I could lay out that I could not set points for something to be built in mid air because the screen on computer showed it. It was a learning experience for all of us getting the verbiage down and understanding of like terms meaning something differently. But I had fun. A lot of head scratching and grey hair and loss of hair but we all seemed to figure it out.
I'm thinking about what sort of workflow you might use that would still have some productivity. I think you would probably set out a number of survey tripods, in scan positions, then measure them with the total station from at least two positions, while someone else operates the scanner, then you'd leapfrog the tripods to the front once the scanner has occupied them. Finding the right number would depend on the efficiency of the workflow.
I find it's too time consuming to mark out all the positions on the ground, with the GLS we usually resect our positions off prisms and measure their occupations as well with the total station or another scanner. We use a Leica height hook when we measure occupation height (and photograph), but we usually prefer to use fixed height prisms with resection for productivity reasons. With the GLS we try to limit prism scans to 3 per occupation, as each takes about 1.5 minutes to complete, while the SX12 can perform rounds of multiple prisms in the same time frame. This time is allowing for time to enter the prism target id into the scanner, the scanner to find the prism, then scan it, until it is ready for the next position. We don't enter coordinates into the GLS, coordinates are calculated later using LSA and entered into registration software.
Topcon's standard workflow is to measure all the occupation points first after measuring them with a total station, enter them into the scanner then carry out a scan traverse, but we don't use this workflow as it's too slow, as it requires marking points on the ground and measuring occupation heights etc.
With the P40, you'd perform a full dome scan and scan each target using a window scan, I recall the P40 has a traverse style workflow. I don't know whether you can get away without entering coordinates into the scanner when using the traverse workflow. I suspect you can just make window scans of the targets and provide coordinates for targets and occupation later in Cyclone Core, that would be how I would do it if possible. The total station can occupy the same tripods as the scanner.
We've found that productivity is improved with more than one scanner, we typically use at least two GLS and one or two SX12's. We typically shoot forward tripod positions before occupying them, so this allows us to get two or more shots of that position before it's occupied and then perform 3 target shots from it after occupying, so we've got shots going to and from at similar times to identify any vertical refraction.
We had a job recently where we worked in geodetic coordinates, scanning a dragline house, the machine isn't level and the scans had to be on machine coordinates, so we've taken some levels off machined pads on the machinery deck, then we transformed our control into machine coordinates using vector rotations, using points and levels measured on the machine. We had to then transform the scans as the instruments were levelled, but the machine wasn't, happy days.
We had a job recently where we worked in geodetic coordinates, scanning a dragline house, the machine isn't level and the scans had to be on machine coordinates, so we've taken some levels off machined pads on the machinery deck, then we transformed our control into machine coordinates using vector rotations, using points and levels measured on the machine. We had to then transform the scans as the instruments were levelled, but the machine wasn't, happy days.
Your local Trimble dealer will happily sell you an Object Oriented Setup option for Access
And there is one for LAS export too
They really help with that sort of malarkey
@pfirmst oh those are fun days. Now you can use N E and UP coords as a work around and was not a bad way to work as datum was honored to a degree and the machines items were as well. Just 0,0,0 was not always were the mechanical engineers wanted it but ended up working pretty well. I used gps to get on datum never cared about state plane really and just used N E and UP coords a lot
@jimcox yeah I believe honestly it was the Navy that pushed for that option as I was involved with that as we were getting ready to purchase the sx10 at the time. The measure a plane and a few more things we needed at times. Yeah that object orientation would work for sure.