trouble refining the registration with target spheres

Discuss FARO SCENE software here.
fobos8
V.I.P Member
V.I.P Member
Posts: 246
Joined: Sat Jul 07, 2018 9:19 pm
5
Full Name: Andrew
Company Details: NDC Surveys
Company Position Title: Surveyor
Country: Uk
Linkedin Profile: No
Has thanked: 56 times
Been thanked: 22 times

Re: trouble refining the registration with target spheres

Post by fobos8 »

Steve, I'd reach out to your Faro contact, send someone at Faro the scans and get them to see what they can do with the data.

I've found them incredibly helpful in the UK. Your resolution of 1/4 should be fine for spheres 15 metres away.

Is there a lot a glass on the building - look like there might be on the first picture? This will play havoc with the laser.

At the bottom of the cloud to cloud screen there is an option to calculate target based statistics. If you tick it it will calc stats on the spheres you have.
skybornevisions
I have made 10-20 posts
I have made 10-20 posts
Posts: 11
Joined: Fri Jul 10, 2020 5:13 pm
3
Full Name: Steve
Company Details: KSC
Company Position Title: CAD
Country: USA
Linkedin Profile: No
Has thanked: 2 times
Been thanked: 1 time

Re: trouble refining the registration with target spheres

Post by skybornevisions »

MajorDomo wrote: Sun Jul 12, 2020 10:41 am Also, is your scanner calibrated?

X330 seem to suffer from decreasing angular accuracy as time goes by, leading to scans to be in essence skewed.
One way to detect this is by placing all your spheres around you, and scan, lift the whole scanner, and rotate by 45 degrees, repeat a scan, and repeat the process a few more times, if you have a hardware/calibration issue you should see some of the scans out of shape pretty quickly.
Interesting, I don't think the scanner has been calibrated since it was bought 6 years ago (but barely used till I showed up)... though the Faro sales rep that gave me a crash course over 6 months ago admitted that the X330 inclinometer was known for not being very accurate. I had noticed the scanner's firmware hadn't been updated since purchase, and some of those updates included improving the inclinometer readings; but after updating to the latest firmware it wasn't obvious that any improvements had happened. I did notice that there were times when the inclinometer readings appeared frozen after moving the scanner to the next position. I wouldn't be surprised if that effected my scans (before I noticed that). Since doing a factory reset that issue has mostly gone away; but when I do notice that inclinometer sensor seems frozen I simply reboot the scanner and it seems to start working again.

At what point do you look to see if scans are out of shape? Right after processing, since all the scans are taken from the same point? Or, after initial registration with target spheres? Can you clarify what specifically you'd be looking for? Sounds like a great experiment I'd like to run today...

Thanks MajorDomo
skybornevisions
I have made 10-20 posts
I have made 10-20 posts
Posts: 11
Joined: Fri Jul 10, 2020 5:13 pm
3
Full Name: Steve
Company Details: KSC
Company Position Title: CAD
Country: USA
Linkedin Profile: No
Has thanked: 2 times
Been thanked: 1 time

Re: trouble refining the registration with target spheres

Post by skybornevisions »

fobos8 wrote: Sun Jul 12, 2020 4:53 pm Steve, I'd reach out to your Faro contact, send someone at Faro the scans and get them to see what they can do with the data.

I've found them incredibly helpful in the UK. Your resolution of 1/4 should be fine for spheres 15 metres away.

Is there a lot a glass on the building - look like there might be on the first picture? This will play havoc with the laser.

At the bottom of the cloud to cloud screen there is an option to calculate target based statistics. If you tick it it will calc stats on the spheres you have.
There wasn't much glass on this building actually--other than a small curtainwall (continuous wall of windows) at the east entrance... I've dealt with buildings where I was scanning on the same level as curtainwall... in fact that previous project I referenced "Nolan Idea Bldg" had curtainwall both below and above my target surfaces. However, the last Faro rep I talked to pointed out that one of the many advantages of using target spheres, is reflections don't matter... unless of course you ultimately end up using c2c.

I did try reaching out to the more knowledgeable Faro rep (of the two I know) last week, but so far he hasn't responded--hence my attempt at seeking help here. I'll probably try to bug him again tomorrow. In any case I'm supposed to be done with everything in three days (including having each unique pre-folded panel encoded ready to be machined on CNC table router--the architect had fun making every aluminum composite panel offset and at arbitrary different sizes from one another--including wrapping inside recessed angled and tapered window terminations), and so I've started pressing forward minus two of the scans that seemed to be uncooperative. There's no way at this point I'll finish by deadline; despite my plan of working 16 hour days on this project.... so I'm a little stressed out at the moment; but I ultimately would like a solution to my current issue so I'll know how to handle this problem should it come up in the future.
MajorDomo
V.I.P Member
V.I.P Member
Posts: 205
Joined: Mon Feb 25, 2019 4:42 am
5
Full Name: Major Domo
Company Details: VeritasVfx
Company Position Title: ceo
Country: Portugal
Linkedin Profile: No
Has thanked: 17 times
Been thanked: 53 times

Re: trouble refining the registration with target spheres

Post by MajorDomo »

skybornevisions wrote: Sun Jul 12, 2020 5:03 pm At what point do you look to see if scans are out of shape? Right after processing, since all the scans are taken from the same point? Or, after initial registration with target spheres? Can you clarify what specifically you'd be looking for? Sounds like a great experiment I'd like to run today...
Thanks MajorDomo
When you register, your result should be quite bad, considering the distance you're working at, with vertical deltas being quite high.
Also try to slice vertically and you should find a few different versions of the same ceiling/floor. It will be more visible on distant parts of the scans.

6 years with no calibration and doing cladding work is not something I would jump into.
jedfrechette
V.I.P Member
V.I.P Member
Posts: 1237
Joined: Mon Jan 04, 2010 7:51 pm
14
Full Name: Jed Frechette
Company Details: Lidar Guys
Company Position Title: CEO and Lidar Supervisor
Country: USA
Linkedin Profile: Yes
Location: Albuquerque, NM
Has thanked: 62 times
Been thanked: 219 times
Contact:

Re: trouble refining the registration with target spheres

Post by jedfrechette »

This may or may not be relevant to your specific problems, but at the end of last year we started noticing what appeared to be systematic errors in the target based alignments we were performing in Scene. In particular, vertical residuals were often on the order of 2x horizontal residuals when the inclinometer was enabled during registration. Disabling the inclinometer, would result in much more uniform residuals, suggesting that the scan geometry itself was not distorted. To me this indicated that there was an issue with either the inclinometer measurements or how they were being handled by the software. The symptoms were observed with at least half a dozen, mostly S series scanners, although X series scanners also exhibited the same behavior. This included new scanners and scanners coming straight from factory calibration. None of the scanners we looked at data from seemed to be exempt.

After doing our own testing and going back and forth with Faro for 6 months the Faro hardware guys were convinced that everything was within spec. I was never able to speak with anyone on the software team directly, but early on there was an indication from one of our Faro contacts that there might be an issue with how much weight was assigned to the inclinometer during target based registration. Later after supplying more data and test results I tried to get confirmation if this was still considered to be an issue, but never received a response.

I'm still suspicious that there is a bug in how Faro is handling their inclinometer measurements, so we worked on setting up automatic logging to start tracking inclinometer performance across all of our projects. Unfortunately, this was about the time everything started shutting down for COVID so we haven't gathered much data to date.

We did, however, go back and manually examine Scene's reported Inclinometer Mismatch values for about 600 scans spread across ~45 different projects with a S350 (265 scans) and X330 (326 scans). All projects were aligned with cloud to cloud and we intentionally focused on small (<50 scan) indoor projects so that the alignments should generally be "good". The median inclinometer mismatch for the x330 scans was 0.0307 degrees or 2.04 x the nominal 1 sigma accuracy of the inclination sensor. The median mismatch for the S350 was 0.0230 or 4.34x the nominal 1 sigma accuracy. Both of these are larger than I would expect and correspond to a deviation of ~4-5 mm at 10 meters. What I find particularly suspicious about these results is that the inclinometer for the S350 and X330 performed basically the same, even though on paper the S350 should be about 3 times better. The scanners have different hardware and firmware so the common denominator is processing in SCENE.

Just poking around at the data files there are also some things that start seeming weird in this context. For example, if you use the SDK to crack open the project file and look at the raw InclinometerScanAxis for each scan, you'll see that it's a 3D vector, which is sensible enough. Strangely though, it isn't a unit vector and the third component is always 0. Maybe this is to be expected based on how the system was designed, but if I was debugging my own code it's the kind of result that would start raising red flags immediately.

Long story-short, a lot is going on under the hood to give you that simplified registration report in Scene. Because of how the software is setup it can be difficult to separate out how each of the factors involved is affecting your results so tracking down issues like this can end up being a challenge and ultimately you need to trust that each of the hidden pieces is working correctly. The best advice I can give is to try and keep your SCENE registrations as simple as possible to match the sophistication of the tools. It seems like a more sophisticated constraint system (C2C+Targets+Sensors) would help in your situation and it's been something I've wanted to see Faro implement for a long time. For many of the same reasons you mentioned in your first post I don't find it efficient to try and optimize alignments in SCENE and find it easier to do that in third party software.
Jed
skybornevisions
I have made 10-20 posts
I have made 10-20 posts
Posts: 11
Joined: Fri Jul 10, 2020 5:13 pm
3
Full Name: Steve
Company Details: KSC
Company Position Title: CAD
Country: USA
Linkedin Profile: No
Has thanked: 2 times
Been thanked: 1 time

Re: trouble refining the registration with target spheres

Post by skybornevisions »

jedfrechette wrote: Mon Jul 13, 2020 8:18 pm For many of the same reasons you mentioned in your first post I don't find it efficient to try and optimize alignments in SCENE and find it easier to do that in third party software.
Thanks Jed for the detailed observation! What software would you recommend for processing and registration over Scene?
jedfrechette
V.I.P Member
V.I.P Member
Posts: 1237
Joined: Mon Jan 04, 2010 7:51 pm
14
Full Name: Jed Frechette
Company Details: Lidar Guys
Company Position Title: CEO and Lidar Supervisor
Country: USA
Linkedin Profile: Yes
Location: Albuquerque, NM
Has thanked: 62 times
Been thanked: 219 times
Contact:

Re: trouble refining the registration with target spheres

Post by jedfrechette »

We finalize all of our registrations in PolyWorks, but it has its own set of limitations. Cyclone has always been a good powerful choice too. If you can navigate their maze of different product offerings I think Leica has been doing some interesting things in general recently. Hopefully they'll continue listening to feedback and steadily improving.
Jed
INV_Scanning
I have made 40-50 posts
I have made 40-50 posts
Posts: 46
Joined: Mon Nov 11, 2019 10:18 am
4
Full Name: Sebastian Winkler
Company Details: INVERS Industrievermessung und Systeme
Company Position Title: Project Engineer
Country: Germany
Linkedin Profile: Yes
Location: Essen, Germany
Has thanked: 5 times
Been thanked: 6 times

Re: trouble refining the registration with target spheres

Post by INV_Scanning »

Hi Steve,

did you try to leave out the inclinometer data of the two scans for registration.

SCENE sets different weightings to different information. A simple target point (Sphere, Checkerboard, point) has a weighting of 1. The inclinometer data is weighted with a value of 1000 (when i looked it up last time).

With badly taken inclinometer data, which can happen, you would just see that your normal targets are with high differences in those two scans.

Maybe this helps!

Regards Sebastian
User avatar
bananmac
I have made 40-50 posts
I have made 40-50 posts
Posts: 48
Joined: Thu Nov 13, 2014 10:58 am
9
Full Name: Maciej Banaszkiewicz
Company Details: Schenkel Vermessungen AG
Company Position Title: Laser Scanning Specialist
Country: Switzerland
Linkedin Profile: Yes
Has thanked: 4 times
Been thanked: 3 times

Re: trouble refining the registration with target spheres

Post by bananmac »

skybornevisions wrote: Sun Jul 12, 2020 5:03 pm
Interesting, I don't think the scanner has been calibrated since it was bought 6 years ago
Hi Steve!

After 6 years I would definetely start from a proper calibration. You can try to do lots of magic tricks with the data but eventualy if scanner is source of the problem, nothing else but calibration will help...

If you need to be 99% sure that your registration is accurate, you should always measure reference points with total station! Otherwise when an error comes up it will be just a guessing game...
skybornevisions
I have made 10-20 posts
I have made 10-20 posts
Posts: 11
Joined: Fri Jul 10, 2020 5:13 pm
3
Full Name: Steve
Company Details: KSC
Company Position Title: CAD
Country: USA
Linkedin Profile: No
Has thanked: 2 times
Been thanked: 1 time

Re: trouble refining the registration with target spheres

Post by skybornevisions »

INV_Scanning wrote: Mon Jul 20, 2020 6:16 pm Hi Steve,

did you try to leave out the inclinometer data of the two scans for registration.

SCENE sets different weightings to different information. A simple target point (Sphere, Checkerboard, point) has a weighting of 1. The inclinometer data is weighted with a value of 1000 (when i looked it up last time).

With badly taken inclinometer data, which can happen, you would just see that your normal targets are with high differences in those two scans.

Maybe this helps!

Regards Sebastian
Thanks Sebastian. Yes, I did try leaving out the inclinometer data on those two scans. Seemed to help at first till I tried decreasing distance error again; but perhaps I was seeking a false "perfection" since everything was in the "green" from the beginning... but yes, disabling those two scans' inclinometer readings did reduce the distance error a little; though I'd find that they would go out of alignment on the other side of where the scan was taken... albeit not as much.
Post Reply

Return to “FARO SCENE”