Taming errors... pt. 6: The Men Who Stare at Points

[Daniel Wujanz]

This part is less technical and more political instead. But don't worry - It'll become very technical ; )

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

B.t.w. It is not my intention to provoke people on purpose. Yet I firmly believe that things are developing in a dangerous direction. Hence, I'd like to point them out at least. It is up to every single engineer to decide how these issues are addressed in practice.

All the best

Daniel

[VXGrid]

[...]
B.t.w. It is not my intention to provoke people on purpose. Yet I firmly believe that things are developing in a dangerous direction. Hence, I'd like to point them out at least. It is up to every single engineer to decide how these issues are addressed in practice.
[...]

Daniel, I see you are not only good in block alignment, but in making enemys as well
They will be out for your blood!

I don't know anymore the exact wording or where I heard or read the sentence, but it was about blaming the sales persons, that your laser scanner has 1mm accuracy, Registration with "our" (manufacturer) software is always working 100% correctly (see we did two scans and in the report this number says 1 mm).

There are two more aspects I like to consider (warning: across the border- people clustering):
1. Practice makes perfect, but practice without someone who can point out issues (teacher) won't help to improve. Since there are a lot of people coming into laser scanning who are masters of other professions (like carpenters, architects, construction engineers, ...) who merely use laser scanning as a tool, they might not think about survey-engineering aspects necessary.
2. Given the different reports, numbers, matrixes, wording and varying scales of numbers of different registration software packages, which might look alike but mean completely different things, people are simply confused. Yes if the numbers for the accuracy are all 0mm deviation, of course, this is alright, but what are good or bad numbers and how are they affecting each other?

I mean you mentioned the issues yourself in your second part of the series: Looking at the registered data should be done when using IPC, to suspend logic errors, otherwise you end up like your image 6 of said article:

A double check nevery hurts, it just shouldn't be the only quality measure one does and rather be a small second check.


I think I just came up with a game while writing this:
We all upload a registration report of a project with 10 scans and other people have to comment what the numbers, tables, matrixes and colours mean and if the project was registered well.
Somebody in?

[Daniel Wujanz]

Dear Martin,

I absolutely agree! The whole motivation of this series was and still is to educate people who haven't got a surveying background (or simply forgotten everything they've once learned). There's no university degree "Laser scanning engineer" or protected profession yet, so everybody can be part of this industry. Yet, it should be in the interest of our industry to establish and satisfy certain standards.

We've heard of a couple of cases here in Germany where clients say "no more laser scanning - use a total station next time" since people repeatedly screwed up.

Regarding the series:

Yes, there will be a lot of practical examples for every single quality measure which will be introduced. I've actually offered some colleagues in the forum helping to find errors in their networks but didn't get any response.

And yes again, you can send me data that we can analyse and optionally discuss it in the series.

And another yes (since you've gently asked), you can definitely help me with all that work ; )

All the best and have a good weekend

Daniel

[Scott.Warren]

We all upload a registration report of a project with 10 scans and other people have to comment what the numbers, tables, matrixes and colours mean and if the project was registered well.
Somebody in?

Put a 10 scan raw data-set online, get us all to register the same data and kick out a report, and see the differences? Interesting.

[Daniel Wujanz]

What about publicly available sample datasets? Such as:

https://shop.leica-geosystems.com/blk36 ... -downloads

The indoor dataset looks suitable since it contains some options for closing loops.

Any other suggestions?

Cheers

Daniel

[VXGrid]

We all upload a registration report of a project with 10 scans and other people have to comment what the numbers, tables, matrixes and colours mean and if the project was registered well.
Somebody in?

Put a 10 scan raw data-set online, get us all to register the same data and kick out a report, and see the differences? Interesting.

Exactly.
But the interesting part of the comparison are not the values itself, but the meaning behind the values.
When comparing values from different reports I fear we will compare apples and oranges...

What about publicly available sample datasets? Such as:

https://shop.leica-geosystems.com/blk36 ... -downloads

The indoor dataset looks suitable since it contains some options for closing loops.

Any other suggestions?

Cheers

Daniel

The top right one or the left bottom one?

Faro raw data can be processed in Register360 and Cyclone with Inclinometer data right?

[Daniel Wujanz]

Dear Martin,

see attachment. This dataset contains 21 scans but should be still fine.

All the best

Daniel

[stutosney]

We all upload a registration report of a project with 10 scans and other people have to comment what the numbers, tables, matrixes and colours mean and if the project was registered well.
Somebody in?

Put a 10 scan raw data-set online, get us all to register the same data and kick out a report, and see the differences? Interesting.

That would be a pretty interesting article, but wouldn't we just end up with different interpretations of what the numbers mean based on peoples different opinions and experiences? The numbers don't always tell the full story!

[Daniel Wujanz]

Dear Stuart,

I reckon that's the question - knowing what numbers mean and what they are based on.

The numbers depend on the software you're using, so that's a good opportunity to revisit the owners manual ; )

Cheers

Daniel

[Daniel Wujanz]

Since I didn't hear any other suggestions regarding which data we should use, I've picked the public data that I've suggested. I will produce and discuss heaps of numbers and what they tell you based on this and other examples throughout the series.

Throwing numbers at people is one things, letting the numbers speak and tell you what they say is another and takes some time...

Cheers

Daniel