Taming errors... pt. 11: How to be precisely imprecise?

Please post any further education or courses that are for laser scanning
Post Reply
User avatar
Daniel Wujanz
V.I.P Member
V.I.P Member
Posts: 177
Joined: Fri Aug 24, 2018 11:26 am
4
Full Name: Daniel Wujanz
Company Details: technet GmbH
Company Position Title: 3D Laser Scanning Specialist
Country: Germany
Linkedin Profile: Yes
Location: Berlin
Has thanked: 161 times
Been thanked: 108 times
Contact:

Taming errors... pt. 11: How to be precisely imprecise?

Post by Daniel Wujanz »

Dear all,

here's issue 11 of taming errors. This time it's all about uncertainty propagation.

https://www.linkedin.com/pulse/taming-e ... iel-wujanz

All the best

Daniel
User avatar
smacl
Global Moderator
Global Moderator
Posts: 1074
Joined: Tue Jan 25, 2011 5:12 pm
11
Full Name: Shane MacLaughlin
Company Details: Atlas Computers Ltd
Company Position Title: Managing Director
Country: Ireland
Linkedin Profile: Yes
Location: Ireland
Has thanked: 457 times
Been thanked: 466 times
Contact:

Re: Taming errors... pt. 11: How to be precisely imprecise?

Post by smacl »

Great addition to the series Daniel, would you believe I have 3 separate editions of Ghilani on the shelf behind me along with two editions of Schofield's 'Engineering Surveying' which are where I look when I'm scratching my head about control problems. I'm a fan of looking at error ellipses to see where things are going wrong on a constrained network as they also hint at which component of the observations are at fault (e.g. angles or distances, more often angles) and provide good basis for the argument to include gyro obs when tunnelling. We do a lot of work on wriggle (circularity of section) analysis to determine deviation between construction and design of cylindrical tunnels where tolerances are tight and decent control is scarce.
In order to stabilise the given network, you could simply add four more control points to the other end of the tunnel so that it complies with the configuration in Figure 4.
In order to stabilise the middle of network b we simply add two more control points close to its centre.
You need to be a bit careful here perhaps as to the provenance of these points, there is a temptation to consider them fixed whereas in truth your control points will also have an amount of uncertainty based on how they were observed and adjusted and will need their own constraint equations. I'd usually recommend a minimally constrained adjustment prior to the constrained adjustment to check the integrity of observations independent of control. While the station values might by rubbish, if you have significant redundancy in the network (e.g. cross-braces, gyro obs etc...) you can get a strong indication that your measurements are sound. That extra control point in the middle of the tunnel is often likely to be dubious :D

Great to see rigorous survey methodology being applied to scanning registration, really looking forward to the next in the series.
User avatar
Daniel Wujanz
V.I.P Member
V.I.P Member
Posts: 177
Joined: Fri Aug 24, 2018 11:26 am
4
Full Name: Daniel Wujanz
Company Details: technet GmbH
Company Position Title: 3D Laser Scanning Specialist
Country: Germany
Linkedin Profile: Yes
Location: Berlin
Has thanked: 161 times
Been thanked: 108 times
Contact:

Re: Taming errors... pt. 11: How to be precisely imprecise?

Post by Daniel Wujanz »

Dear Shane,

thanks heaps for your great comment - solid gold : )

I'd like to respond to one aspect which will be subject in the next episode, namely weight / precision / uncertainty (depending on how old you are).

smacl wrote: Wed Nov 16, 2022 1:52 pm You need to be a bit careful here perhaps as to the provenance of these points, there is a temptation to consider them fixed whereas in truth your control points will also have an amount of uncertainty based on how they were observed and adjusted and will need their own constraint equations. ...
Great to see rigorous survey methodology being applied to scanning registration, really looking forward to the next in the series.
As for our software (and the theory of surveying) NOTHING is god-given and error-free. That means that everything can receive individual variance-covariance information (aka weights) - let's say tacheometric control points receive accuracies around 2 mm and GPS points have values of 2 cm horizontally and 3 cm vertically. B.t.w.: What people tend to forget is the error budget of the local target's centre which is then tied to a superior coordinate.

I don't want to say too much, but apart from assigning weights, it is even moren important to check if your assumptions were satisfied. All fun and games : )

All the best

Daniel

P.S. I just have one edition of Wolf / Ghilani. Since the founder of the company was a professor for adjustment calculus and engineering geodesy we mostly use our own stuff.
User avatar
smacl
Global Moderator
Global Moderator
Posts: 1074
Joined: Tue Jan 25, 2011 5:12 pm
11
Full Name: Shane MacLaughlin
Company Details: Atlas Computers Ltd
Company Position Title: Managing Director
Country: Ireland
Linkedin Profile: Yes
Location: Ireland
Has thanked: 457 times
Been thanked: 466 times
Contact:

Re: Taming errors... pt. 11: How to be precisely imprecise?

Post by smacl »

Daniel Wujanz wrote: Wed Nov 16, 2022 2:16 pmP.S. I just have one edition of Wolf / Ghilani. Since the founder of the company was a professor for adjustment calculus and engineering geodesy we mostly use our own stuff.
Schofield is also worth having on the shelf as he covers a few things not in Ghilani & Wolf, notably relative error ellipses on lines. Nice to have a professor for adjustment calculus and engineering geodesy to cover this stuff. I struggle with it and grab hold of all the resources I can get my hands on.
Post Reply

Return to “Education & Courses”