NavVis VLX 3 vs. TLS: Round 2 - You be the judge
HTML-код
- Опубликовано: 18 сен 2024
- It's time for Round 2 🥊 and we want you to be the judge.
Download the point cloud data here to see for yourself 👀 navv.is/43wmPiX
#BuildBetterReality #TLS #LaserScanning
This is misleading as you didn't show the time needed to set up control points for VLX scanner (or they just magically created by themselves?). TLS systems are usually accurate enough to rely on cloud to cloud only, therefore no total station needed inside. According to our estimations efficiency are roughly the same between RTC360 (cloud to cloud) vs VLX+totalstation.
I think your guess is quite reasonable. However, if you observe the walkthrough process in the point cloud carefully, you will find that control points are not used at all. To a large extent, this indicates that the point cloud registration algorithm of NavVis is very efficient in extracting point cloud features (though this is just my inference). If we assume such an efficient registration algorithm, then the scanning efficiency of NavVis must far exceed that of TLS technology. However, if we talk about accuracy, as a type of Mobile Laser Scanning (MLS), it is difficult for NavVis to compete with the accuracy of static scanning in TLS, because the process of MLS inherently involves stacking multiple frames into one state. No matter how accurancy the algorithm is, even the slightest registration error will be amplified over multiple scans.
@@leonzhu528 control point is scanned at 0:43. Accuracy is a different topic, RTC 360 is not the most accurate scanner and produces quite a bit of noise, so in this regard I prefer VLX. But, due to the drift of the VLX system control points are a necessity. I simply pointed at misleading info from the video. Also there are no need to guess how efficient the algorithm is, if we need to have even a single control point in the middle of the building, that means that we need to do an extra time consuming job with total station. We are in a business for the past ten years, the last thing we want is to use total station with scanning again.
Hi Daniil, for an environment of such a size, control points are really not necessary. They are of great help of course when scanning larger, elongated areas where loop closures are not possible. In this case, we wanted to show how you can pick them with VLX and use them for registration and QC.
By the way: scanning an elongated area with a TLS where you cannot close a loop also benefits from using control if you really want to be in the mm level.
Yes, SLAM systems work quickly and seem to be convenient. But in the algorithm itself there is an accumulation of errors, to compensate for them, you must have closed the loop. Any stationary scanner will give excellent accuracy and repeatability of measurements, but SLAM does not. On this small closed object, this may not be noticeable, but if the object is extended, then you will never get 2 identical point clouds. What sensors are used here, VLP-16? They have an accuracy of about 5 cm, and there can be no talk of global stitching accuracy. The RTS-360 is a great scanner, fast, accurate and compact.
No doubt, the RTC 360 is great (like the X7 or Faro Focus +). To create a point cloud of a site, you need to register each setup to its neighboring setups. In fact, this is exactly what SLAM based system are doing as well, just around 20 times per second. Also, you will close loops when using TLS all the time when you have more than one neighboring setup (sometimes also called bundle adjustment). In the end, the systems are not that different at the core, it's rather the frequency of scans.
Of course, it also depends on the HW that is built in. The VLX 3 doesn't use VLPs anymore but scanners that is specifically designed for the surveying industry with 32 lines and 40° opening angle creating more than 1.3 Million points per second.
Having that said, check out the point cloud download of the VLX 2 that uses VLP-16s. The accuracy is just stunning (6mm 1sigma locally) and used by hundreds of surveying professionals around the globe. The difference is on the software.
At some point you decided to buy a TLS (I assume you started with a total station?). Now it is about time to add a new tool to your belt in addition to TLS.
All these arguments about the accuracy of Slam systems are shattered when you measure roads, bridges, subway tunnels. Even with a simple alignment of two separate surveys. You will never get the accuracy of a TLS scanner. What is 6mm + sigma (LOCALLY)? this essentially means that at every point in time you are adding a 6mm error to your measurements. if the algorithm does not have enough surfaces for consistent adjustment, you risk getting incredible artifacts, regardless of the built-in gps.
What about of GLOBBALLY accuracy?
The TRC 360 also has an inertial system, like some other scanners, so there is definitely no problem with alignment of stations.