Looking for cheap facility 3D scanning solution

Joe Dunfee

After receiving very poorly done and inaccurate drawings from an architect, and a customer that put off giving us physical measurements for two months, I am now doing a lot of re-design on my drawings.  Most of the main dimensions were off from 1 to 2 feet.

I have long been hoping for a way to get the real world facilities into my computer with affordable equipment. I had hoped that Google's Project Tango was the solution.  The idea is that I would mail a Tango device to the customer, and have him create a 3D scan of the facility. The Tango device would be consumer friendly, so the customer could learn to do it. Plus, the "Cool factor" may be enough to motivate them to do it. And it would be cheap enough that I would be willing to risk having it in the hands of a customer. And if they were not quick to return the device, I could have a few spares.

But, I spoke to to a vendor about the Tango. He says if I attempted to use the device for a warehouse, it would really not be accurate enough. It would only be +/- 6", and then the registration of several scans would be much worse.  Also, it would not be able to scan things like trusses and small pipes, which would be important to us.

I have noted for a decade now, that for under $100, I can purchase a laser distance measurement device capable of 1/16" accuracy and 100' range. I realize these are not capable of fast measurements, but I am puzzled why I can't buy a $200 room scanner that I can sit in the middle of a facility and let it run overnight.  Or alternatively, the type of scanner that puts out a fanned-laser and a camera to see how the line distorts.  Very high resolution cameras are very cheap now, and that would be an even better solution in terms of the speed and resolution.

My guess is that if there were a low-cost solution (i.e. under $500 USD), I imagine it would be very popular, and I would have already found it. So, perhaps there are some  technical issues with the low-cost methods that I am ignorant of.  However, it may also be that the idea of a cheap, but very slow scanner, or one that could only work at night, were just not marketable.

Is there a solution out there that I am unaware of?

Norbert Probst

Hi Joe
I've read (hard copy) in Racetech magazine that a simple Wii receiver is
on of the best scanners available. I have not tried it but I believe that the magazine
has an online presence. They build a space frame around the object to get a better
scan/cloud.... lawn mower style... as it were. Easier to stitch later.
check it out.
Regards
Norb

Joe Dunfee

Neither "Scanner" nor "Wii" in their search tool shows anything. But, aren't you referring to the Kinect with its Primesense technology?

When I have spoken with people who sell scanners, they said the Kinect type of scanner will not work at the warehouse scale because it is not precise enough. The start and end scans don't end up getting registered to each other, and end up far apart. From my own examination of the Kinect specs, I can see that the resolution is not dense enough to see things like roof trusses. And even if the 16ft distance were acceptable (the ceilings in many of the warehouses are often over 25ft.) 7 cm horizontal/vertical resolution at the maximum range is not even close to being able to see the roof trusses well enough to model them. The depth accuracy Kinect's depth accuracy is 2.75" at that range.

Yes, there are some software packages that can use the Primesense type of sensor and generate a pretty picture of a room. But, even on that smaller scale, don't expect to be able to measure a scanned window in that room, to determine the size of a window blind you want to purchase.

A student at a university will do things like create a laser based scanner that would have the specs we need to do a warehouse scan, using only a few hundred dollars of equipment. The specs seem to be sufficient to do at least a small warehouse. But, those low-cost projects never turn into commercial products. Perhaps there are technical limitations that the student does not write about in their paper, and it is really not viable.

-Joe

Hosepo

Kinect... You know, I've tried it. I've tried it with free software, payed software, professional software! I'll show it all here: Windows native application, Scanect, Artec Studio. The result was pretty same - rubbish. The resolution is too low, You can capture just general shape, but nothing else.
I'm sorry, I was quite emotional..
The prefect cheap way to scan is to use photogrammetry instead of Kinect. Just a camera, good smooth lightning, a tripod (to prevent shakes) and a special software. They all give pretty same result just use whatever seems more suitable for you. The main idea is to make photos from different angles. From all angles actually Than the algorythm calculates 3d shape from the photos. As I know white-lite 3d scanner (not laser ones) uses pretty same technology.

Tom Foster

I'm hoping and dreaming that Brics will have something to announce about point cloud integration at the conference, and that this will not just be about laser scans but will extend to photogrammetric input.

i.e. receiving meshes maybe even photorealistic textured meshes from other programs e.g. my favourite https://www.capturingreality.com - without getting indigestion with the large datasets involved. I doubt that Brics will want to restrict itself to the Adesk offerings in that field.

Hints at the last conference suggested that the Brics solution will somehow be based on or integrated with VR. That would put latecomer Brics ahead of the field.

Joe Dunfee

@Hosepo said:
I'm sorry, I was quite emotional..

You've been fooled by the marketing department of several companies, and they have successfully wasted many hours of your time. Hopefully, you didn't spend any money that didn't get refunded.

-Joe

Joe Dunfee

As I re-read this thread before commenting, I was reminded of something I posed. There exist on the market, relatively cheap laser distance meters. It may take them a few seconds to get a good reading, but they are quite accurate at typical distances needed in industrial facilities. A local hardware store has ones starting at $20 USD for a model that will measure 50 feet, +/- 1/8". I just did a search for Time-of-Flight cameras, that do multiple pixels, and see that now there are many of them on the market. Though, the few I looked at didn't have the accuracy I would want for measuring a building.

Many in the world of drones are very interested in those kinds of 3D sensors, for their ability to map the environment for the drone for navigation. They are also interested in their ability to do measurements of things like grounds for excavation, or grain silos.

This is a VERY rapidly changing field. So, I am hoping to eventually see a good, fully integrated, solution soon, that has useful resolution and accuracy. Perhaps they will soon start to approach what the marketing departments have been trying to sell for several years!

-Joe

Richard Sands

Joe those little hand held distance measurers are pretty awesome.
I bought a $30 one and tested it against my Leica.
I couldn't fault it.
Obviously these measures still need care in use. And held horizontal unless it's got an inbuilt slope correction.

I attended a seminar the other day and a bloke was talking insurance claims and some scanners that people had taken as gospel, without recourse of checks. And more checks.
It seems they're not all they are made out to be.