I have had this issue a few times on different projects and was wondering if anyone else has experienced this:
After completing a building scan, I hammer in a few nails around the perimeter, get the coordinates of the points using GNSS Receiver, then place a checkerboard target accurately over the nail and scan the target.
Back in RW, after applying the coordinates to the targets I will get good averaging error:
However, when I check the cloud, there is definite tilt.
BEFORE GPS (Level using scanner Inclinometer)
AFTER GPS
Any input would be greatly appreciated
Pointcloud Tilting after GPS Adjustment
-
- V.I.P Member
- Posts: 202
- Joined: Sun Sep 13, 2015 7:19 pm
- 8
- Full Name: Jared
- Company Details: Sova Surveys Limited
- Company Position Title: Director
- Country: UK
- Skype Name: jaredmatthews
- Linkedin Profile: Yes
- Has thanked: 5 times
- Been thanked: 13 times
- Contact:
Pointcloud Tilting after GPS Adjustment
You do not have the required permissions to view the files attached to this post.
- smacl
- Global Moderator
- Posts: 1409
- Joined: Tue Jan 25, 2011 5:12 pm
- 13
- Full Name: Shane MacLaughlin
- Company Details: Atlas Computers Ltd
- Company Position Title: Managing Director
- Country: Ireland
- Linkedin Profile: Yes
- Location: Ireland
- Has thanked: 627 times
- Been thanked: 657 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
What type of GPS receiver and how long do you occupy the points for? Worth remembering that GPS near buildings can suffer from multi-path error and has accuracy in the order of a few CMs in height. The following paper suggests 1 CM RMSE for static GPS with 30 minutes occupancy in open areas and 4 CM RMSE near buildings. RTK GPS with shorter occupancy times will be worse. For higher accuracy, hold one GPS position fixed, constrain the other positions and traverse and level in your other external control. If the job doesn't warrant this level of effort (very often the case), don't include your GPS in your registration but use it to do a Helmert transformation on the registered point cloud. This will position the point cloud on the GPS grid, to the accuracy of the GPS available without messing with your registration. If you have multiple GPS points but don't want to inherit GPS scale factor (or other related interference factors), do a six parameter / rigid body transform rather than a Helmert or 7 parameter transform.
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
Dear Jared,
in addition what Shane already suggested, you should also check the location of targets, e.g. straight lines of targets are bad or all of them in a relatively small area favours extrapolative effects. If the listed residuals from RealWorks are realistic, the targets appear to be very close to each other...
In order to solve your issue (without knowing the exact target configuration), you just have to decrease the weights of the GCP-components. E.g. assign sigma x / y to 2 cm and sigma z to 6 cm (the height component is always less precise than the locational one). This will shift your network to the desired location without introducing tensions or causing misalignments to the scanning network.
The angular accuracy of the applied inclinometer (you are using a Faro scanner, right?) is about 0.3 mrad. Hence, the rotations about the x and y-axes should not be biased by the suggested a priori weights for the GCP-points.
Or choose the dark side: castrate the functional model (apply a four parameter transformation) and do not rotate about the x and y-axes...
All the best
Daniel
P.S. how do you place the target over the nail? That can also be another minor error component...
in addition what Shane already suggested, you should also check the location of targets, e.g. straight lines of targets are bad or all of them in a relatively small area favours extrapolative effects. If the listed residuals from RealWorks are realistic, the targets appear to be very close to each other...
In order to solve your issue (without knowing the exact target configuration), you just have to decrease the weights of the GCP-components. E.g. assign sigma x / y to 2 cm and sigma z to 6 cm (the height component is always less precise than the locational one). This will shift your network to the desired location without introducing tensions or causing misalignments to the scanning network.
The angular accuracy of the applied inclinometer (you are using a Faro scanner, right?) is about 0.3 mrad. Hence, the rotations about the x and y-axes should not be biased by the suggested a priori weights for the GCP-points.
Or choose the dark side: castrate the functional model (apply a four parameter transformation) and do not rotate about the x and y-axes...
All the best
Daniel
P.S. how do you place the target over the nail? That can also be another minor error component...
- smacl
- Global Moderator
- Posts: 1409
- Joined: Tue Jan 25, 2011 5:12 pm
- 13
- Full Name: Shane MacLaughlin
- Company Details: Atlas Computers Ltd
- Company Position Title: Managing Director
- Country: Ireland
- Linkedin Profile: Yes
- Location: Ireland
- Has thanked: 627 times
- Been thanked: 657 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
+1000 on this. If you're carrying out any kind of transformation or control network the detail (objects your surveying) should lie within the bounding polygon of the control points. You should also avoid short baselines where possible, though this is more of an issue with total stations.Daniel Wujanz wrote: ↑Mon Aug 17, 2020 8:55 amyou should also check the location of targets, e.g. straight lines of targets are bad or all of them in a relatively small area favours extrapolative effects.
Not sure there's a castrate option in realworks...Or choose the dark side: castrate the functional model (apply a four parameter transformation) and do not rotate about the x and y-axes...
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
Touché, Monsieur Shane! Maybe you're right...let's move to plan c:
If you can't do it in RW use any adjustment calculus software to do the transformation (e.g. JAG3D, which is free https://software.applied-geodesy.org/en/). Then you can apply the computed transformation parameters by writing and applying a rmx-file that contains all scans in the project... or maybe there's another option to systematically shift and rotate the project.
All the best
Daniel
-
- V.I.P Member
- Posts: 202
- Joined: Sun Sep 13, 2015 7:19 pm
- 8
- Full Name: Jared
- Company Details: Sova Surveys Limited
- Company Position Title: Director
- Country: UK
- Skype Name: jaredmatthews
- Linkedin Profile: Yes
- Has thanked: 5 times
- Been thanked: 13 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
Thanks for the input Daniel and Shane. I appreciate your suggestions.
Just to clarify my workflow:
Receiver is Trimble R4 - model 3.
Checkerboards are vinyl with a tiny hole in the middle where the intersecting B and W areas meet. This is placed on the ground over the nail.
Scanner is a Faro.
The GPS points were positioned as best as possible given the site situation. It was a residential property, so I took readings at each corner of the building (obviously not right up close to it).
I have had this issue before in the passed. The thing it, it happens on some projects, and on others it doesn't.
The occupancy at each point wasn't time-based, but rather 100 readings that are then averaged out.
My feeling is that it may be a RW cause.
I am not sure there are ways to disregard transformations in certain axis in RW. (ie, not sure you can apply a Helmert transformation on the registered point cloud)
Just to clarify my workflow:
Receiver is Trimble R4 - model 3.
Checkerboards are vinyl with a tiny hole in the middle where the intersecting B and W areas meet. This is placed on the ground over the nail.
Scanner is a Faro.
The GPS points were positioned as best as possible given the site situation. It was a residential property, so I took readings at each corner of the building (obviously not right up close to it).
I have had this issue before in the passed. The thing it, it happens on some projects, and on others it doesn't.
The occupancy at each point wasn't time-based, but rather 100 readings that are then averaged out.
My feeling is that it may be a RW cause.
I am not sure there are ways to disregard transformations in certain axis in RW. (ie, not sure you can apply a Helmert transformation on the registered point cloud)
- smacl
- Global Moderator
- Posts: 1409
- Joined: Tue Jan 25, 2011 5:12 pm
- 13
- Full Name: Shane MacLaughlin
- Company Details: Atlas Computers Ltd
- Company Position Title: Managing Director
- Country: Ireland
- Linkedin Profile: Yes
- Location: Ireland
- Has thanked: 627 times
- Been thanked: 657 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
It could be RW but it is worth remembering that GPS is subject lot of potential interference factors that need to be provided for so I wouldn't be surprised to see differences from one site and/or time to the next. I'd go with Daniel's advice re the RMX file. If you're coming unstuck you're welcome to a demo / loan license of SCC to do the transform but this would necessitate moving your registered cloud in and out of a neutral format such as E57 or LASjaredmatthews wrote: ↑Mon Aug 17, 2020 9:55 amMy feeling is that it may be a RW cause.
I am not sure there are ways to disregard transformations in certain axis in RW. (ie, not sure you can apply a Helmert transformation on the registered point cloud)
Given you're UK based, also worth giving the lads in Korec a shout for best advice re-Trimble kit and software.
- Daniel Wujanz
- V.I.P Member
- Posts: 227
- Joined: Fri Aug 24, 2018 11:26 am
- 5
- Full Name: Daniel Wujanz
- Company Details: technet GmbH
- Company Position Title: 3D Laser Scanning Specialist
- Country: Germany
- Linkedin Profile: Yes
- Location: Berlin
- Has thanked: 206 times
- Been thanked: 188 times
- Contact:
Re: Pointcloud Tilting after GPS Adjustment
Dear Jared,
can you send me the following information in a simple text file?
- local target centre coordinates: file name of all scans that have targets (with GNSS coordinates), target-ID, x, y, z
- superior coordinates: target-ID, X, Y, Z
Shane could process the data in SCC and I would do it in Scantra.
Another issue is of course that your scanner's coordinate system is a planar one while WGS84 is anything but planar...
Cheers
Daniel