PDA

View Full Version : Laser scanning using a distance-sensing laser


normand blais
Thu 03 June 2010, 19:11
Cut from this thread/post (http://www.mechmate.com/forums/showthread.php?p=42793&postcount=12)

...................
Got the link from MagWeb on
http://www.david-laserscanner.com/forum/viewtopic.php?f=8&t=1624&p=12081&hilit=sculptris#p12081 but that is another story...

smreish
Fri 04 June 2010, 17:44
David's Laser works well. I have done many instances of a video camera and laser pointer to create cloud point data. Fun.

Maker Faire or Make magazine is a great resource for this type of DIY laser imaging.

...back to the fun.

Conrado_Navarro
Fri 04 June 2010, 20:08
we also can put the scanner on the Mechmate, so besides a cnc router we can have a cnc scanner. I know it can be done, Just I have no time to do it.

Gerald D
Sat 05 June 2010, 00:38
If you want to mount a simple laser scanner on the MechMate, we can start another thread for that. :)
(there are some simple ideas that do work - simpler than the "David" style)

normand blais
Sat 05 June 2010, 06:20
I have seen some scanner technique on mach forum .but i dont have mach

http://www.machsupport.com/forum/index.php?PHPSESSID=jamtvmk3s0r0ec7r59bv8dkom2&action=search2

Gerald D
Sat 05 June 2010, 09:09
Normand, your link doesn't work

I get the impression that the DIY folk interested in laser scanning are too fixed on the "projected_lines_plus_camera" methods like the "David" method. We folk with CNC tables (or lathes) already have systems for moving a probe to a very precise location. What we actually seek is a non-contact probe than can measure distance to a precise spot.....and lasers can do exactly that.

Imagine having a pin-point beam laser pointing straight down, mounted to our y-car, so that stays a fixed height above the table, movable in the x & y directions. But this is a special laser that has an electrical output of the distance it senses its focus dot to be away from itself. Actually, this concept is well known in devices like this (http://www.stanleytools.com/default.asp?CATEGORY=LASER+MEASURING&TYPE=PRODUCT&PARTNUMBER=77-910&SDesc=TLM+100+FatMax%26%23153%3B+Tru-Laser%26%23153%3B+Distance+Measurer).

So, we need to find a laser distance measuring unit that measures in the range of about 2" to 10", with a resolution of say 0.05", that gives an electrical output that we can log while our CNC motors move the laser across the surface of the object that we are probing. Once we find a cheap enough distance sensing laser, the only other part is the programming. There is a ShopBotter, who was a programmer in a previous life, who built himself a system like this, and now makes money off it. But he doesn't believe in sharing. . . . . .

I have assembled this thread in the hope that we can explore / develop this "spot-height" laser sense method. We need to:
a. find a low-cost laser "head" with output of distance
b. develop programming that works with Mach or EMC that will record the point cloud.

liaoh75
Sat 05 June 2010, 10:21
Gerald, does this shopBotter sell a product we could use?

Gerald D
Sat 05 June 2010, 11:13
No David, he doesn't. And he doesn't discuss his method either.

Here (http://www.cnccontrols.com/laserprobe.asp) is a starting point of a commercially available system.

Gerald D
Sat 05 June 2010, 11:17
PS. I didn't check other forums before starting this thread.....we may be trying to re-invent the wheel. But I doubt that this stuff is available "open-source" for Mach and EMC yet.

normand blais
Sat 05 June 2010, 12:18
That mach link is the search page, try laser scanner and 3d scanner there are video scanning there too
I did the laser line scanning ,my laser been cheap and under power had poor result . I upgraded with a slide projector like they suggest cut a mirror the size of a slide and scratch a line with a razer blade .Nice bright line, thin much better than the laser I had.
The bulb burnt cause the fan stop.Cant find a replacement .So I might look for a super bright led to replace the bulb like in david forum. They even have structured light scanning with micro projector. The annoying thing with david scanner was the calibration pannel .They found a way to calibrate and replay the the line motion so the pannel are not nedded after (some with Mach )

Got the feeling this is the system he use http://www.scantech.dk/3d-laser-scanners/mounted-on-a-cnc.html

David he sell :rolleyes:only to the military

TheDave
Sat 05 June 2010, 19:43
Love this idea!!

Preliminary find:
http://scientificsonline.com/product.asp_Q_pn_E_3095300

I think this does more than what we need and probably costs too much.

Gerald D
Sat 05 June 2010, 20:52
Dave, that laser does not have distance sensing.

Here (http://sensor.baumerelectric.com/productnavigator/scripts/level2.php?pid=&cat=CONFOptoelektronische_Sensoren&ptk=&psg=opProduktgruppe%7C___Distanzsensor&language=en&country=US&header=United%2BStates%2B/%2BBaumer%2BLtd.&restrict=&specialSort=0&headerencoded=United%252BStates%252B%252F%252BBaum er%252BLtd.&selectedPsg%5BopProduktgruppe%5D=___Distanzsensor&orderstring=TM_Opto_Familie%7Casc&order%5BTM_Opto_Familie%5D%5Bposition%5D=1&order%5BTM_Opto_Familie%5D%5Basc%5D=-selected&order%5BTM_Opto_Familie%5D%5Bdir%5D=asc&pageID=1&&orderby=TM_Opto_Gehaeuse|asc) is Baumer's range of distance sensing lasers.

Acuity (http://www.acuitylaser.com/products.shtml) range

It might also be possible to hack into one of these (http://www.laserdistancesensor.us/).

D. van Randen
Sun 06 June 2010, 01:51
Hi Guys , this topic has fascinated me for ages , I have never had the time to build but plenty of time to troll the internet.

Leica has hacked one already

Leica Disto D330i (http://www.leica-geosystems.com/en/Leica-DISTO-D330i_83106.htm)with Blue tooth, resolution is 1/32 or .03mm


there is a very good discussion thread on machsupport.com entitled video p*r*o*b*i*n*g (http://www.machsupport.com/forum/index.php/board,67.0.html). (yes it is spelled that way)

Another word to use when searching laser distance, is displacement. Keyence has several displacement models , but not cheap even if you go through ebay.

Then there is the alternate industrial method using video capture cards like those by Epix, but involves some programming of a camera and laser pointer to create the "blob" for tracking.

or do what this university student did.

http://sites.google.com/site/todddanko/home/webcam_laser_ranger

Gerald D
Sun 06 June 2010, 03:39
I think the trick is to get the "camera" as close as possible to the line of the beam, so that one can probe deep profiles without the camera losing site of the spot. The closer the camera, the more difficult the triangulation. This is why I like the packaged sensors where the camera/detector is in the same housing as the laser emitter and it is already factory calibrated.

hennie
Sun 06 June 2010, 04:18
But they are so expencive:(

normand blais
Sun 06 June 2010, 06:08
Here is a compilation of 3d scanner info and link.http://www.simple3d.com/

Gerald D
Sun 06 June 2010, 08:05
Hennie, you can buy a Baumer distance sensor from RS components (http://za.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=3651651) for R7000 [$1000], which IS expensive, but it is already focused and calibrated and it needs no fiddling around, compared to the David style dot/line/webcam/ DIY setup. And if RS has stuff like this for $1000, it probably means there are reasonable equivalents for much less. And realise that the distance sensor is the ONLY bit of hardware needed to make it work - the rest will be software, or macros for Mach3/EMC which already exist to some extent.

Normand, that is a long list....is there anything in there that helps us to put a Z-distance sensing laser onto an XY table and run it with Mach/EMC?

hennie
Sun 06 June 2010, 09:07
Never thought about it in detail,we are so used to plug and play.

danielpower
Sun 06 June 2010, 09:26
I wonder the price of distance-sensing laser.

PEU
Sun 06 June 2010, 09:48
what about his one: http://www.tormach.com/blog/?p=1038 not laser actually.

Gerald D
Sun 06 June 2010, 12:28
Sorry Pablo, that is 2D scanning only ...... you can call it "tracing", like using old-fashioned tracing paper.

normand blais
Mon 07 June 2010, 05:24
Gerald you right this is a long old link ,lots of missing link .I did not find anythyng . I never tough much about distance sensor scanner, to complicate or expensive for me . There is many type of scanner for many occasions.I am aiming more at scanning living been.dot scanning is too slow for that .Even line scanner like david is at the possible limit with the right equipment. Hacking electronic is to complicate for me I have to wait for some bright guy to guide me as to what to put together. On the local scene here Creaform bougnt Inspect.

Gerald D
Mon 07 June 2010, 05:51
There are distance sense lasers on e-bay for $100 at the moment, but they are the wrong spec. (long distance (low near resolution), big spots) Anybody know what the Leica Disto D330i (http://www.leica-geosystems.com/en/Leica-DISTO-D330i_83106.htm)costs?

D. van Randen
Mon 07 June 2010, 08:48
www.lascolaser.com (http://www.lascolaser.com/product/Leica-DISTO-D330i--Laser-Distance-Meter--With-Bluetooth%AE-_Qitem776748.html) 379.00 usd

Another scanning site , this idea could be quite good speed wise. www.grasp3d.com (http://www.grasp3d.com/)
They also supply lasers for scanning

The Grasp3d site is an offshoot of this site, www.intricad.com (http://www.intricad.com/index.html)

digger
Mon 07 June 2010, 09:11
There is no price tag, but probably could be useful if the price is reasonable.

http://www.aculux.com/Index.htm

riesvantwisk
Mon 07 June 2010, 09:17
This method is very nice :
http://www.geeky-gadgets.com/proforma-3d-scanner-uses-a-standard-web-camera-23-11-2009/

However, I think this system would be bad for relieve scanning.

It seems like each scanning method has it's up's and downs, price, vs speed vs accuracy vs information captured.

Gerald D
Mon 07 June 2010, 09:46
Can I gently point out that the title of this thread is Laser scanning using a distance-sensing laser ?

TheDave
Mon 07 June 2010, 11:35
Everywhere I look, the Leica DISTO™ D330i sells for USD $379. I like the Bosch one you listed, about $100. Definitely worth hacking!

Gerald D
Mon 07 June 2010, 13:07
Before hacking a Bosch, or similar, there are a couple of things that must be checked:
- fairly small spot size
- a good resolution
- and a good speed of getting the data out.

Also realise that the laser/detector can be mounted fairly high up on the y-car, if needs be.

riesvantwisk
Mon 07 June 2010, 15:02
I have used the mentioned sensors (http://za.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=3651651).

I have been developing a system (with some colleagues, I did the software, must be 12 years ago) that could detect if there was cap correctly positioned on a oil drum (50 barrels)
if the cap was in a angle we could detect that, or if it was missing we could detect that.
The system consisted of a stepper motor that would move the laser sensor over the oil drum and measure distances. The output of that was just a single graph where we could see the slope of the top of the oildrum, including the cap, we did see a slope on the cap when it wasn't position right, or a big 'hole' when the cap was missing.

With scanning objects the problem here is that the datastream from the laser sensor needs to be well in sync with the position of the stepper motor, if you want to measure any object with decent speed, you don't want to have a system that goes like 'step->read->step->read', much like the touch probe systems.

With the measurements of the barrels we didn't care to much about this, because we knew the size of a barrel, and thus we could calculate and scale back pretty well, we even didn't care much about the speed of the sensor.
With scanning objects it's required that the software pulsing the stepper motor, and the software that reads the output of the laser distance sensor, must by well synced or else the object appears skewed or mis-formed in some way.

The sensors that RS components are selling (the square version) consist of a simple laser, and a single line CCD array, the CCD array is like found in a good old scanner, the CCD array simply 'detect' the laser dot on one or more of the arrays and based on the angle they know the distance. This is described well here : http://archives.sensorsmag.com/articles/0702/laser/fig1.gif (http://www.eltrotec.com/db/pdf/LDS60e1.pdf)


A DIY version of this could be a single laser pen, and a webcam + red filter. The webcam can then detect the red dot on the object, the Mechmate will then move the camera + laser pen over the object. We simply point the laser down, the object (unless very shiny objects) will have enough diffuse reflection to get spotted by the camera.


Detecting a single dot on a camera image is very simple. You are only limited by scan speed because of the camera that has only (usually) 25 frames/second. The good part is, that a CCD camera has many lines instead of a single array, so you could potentially scan in one scan multiple lines at the same time.
For a better system, you might need two camera's along each side, for optimum laser point detection, may be even two laser in a angle, but that's just optimalisation.


With EMC (http://www.linuxcnc.org/) you have a lot of control and you can add your own code to real time threads if you want to. If you use real distance sensors you then need to have a A/D converter and read that back over serial or parallel port. If you use EMC's real time treads for that then this can be perfectly synchronized so that steppers, position information and hight information can be accurately stored in one file.
If you use a camera stream then somehow this need to be stored on disk, and properly analyzed afterwards, I am not sure if there is enough CPU speed to handle that real time.


With Mach3 I am not sure how much control you have, and if you are allowed to write C code within the Mach3 kernel. if you can, then principles are the same as with EMC. If you cannot then there is not much you can do then step-measure-step-measure sort of system , like a touch probe.
Bottom line is, that if you cannot have your step-pulses synchronized with your read-outs from the sensor, then it's not going to work.

Gerald D
Mon 07 June 2010, 23:49
Ries, that is great experience and an interesting post - thanks!

Laser distance sensing is done by at least 2 different methods:
- the triangulation method (as per the modules mentioned earlier, and Ries's sketch)
- time of flight method (typically longer range, above what we need for CNC)

Both methods employ a beam transmitter with a sensor mounted closely beside it, and therefore they both have a triangle effect. From Ries's sketch above, it can be seen that the sensor(camera) view can be blocked by ridges(hills) standing up from the profile. So, we want to keep the point of the triangle as sharp as possible, to reach down into the details of the "valleys". This is why I like the pre-packaged laser&sensor modules.

(Some commercially available scanning systems have a pair of sensors, one to either side of the laser beam, to increase the chances of the sensor being able to see the spot, but it doesn't totally solve the problem)

My gut feel says it would be nice to mount one of these (http://sensor.baumerelectric.com/productnavigator/scripts/product.php?pid=OADM_13T6480_S35A&cat=CONFOptoelektronische_Sensoren&psg=opProduktgruppe|___Distanzsensor&language=en&r=1&ptk=&country=US&header=United%2BStates%2B%2F%2BBaumer%2BLtd.) modules about 6" above the y-car, so that it reads something on the table that is between 14" to 20" away. The angle at the sense point will only be about 3 degrees.

PEU
Tue 08 June 2010, 04:26
Reading that site I see that measuring distance is adjustable, in one model via a pot, that may mean the depth of focus of these lasers is very shallow which for our use is not a good thing.

Gerald D
Tue 08 June 2010, 04:42
Pablo, I cannot see what you are seeing......which site are you talking about?

PEU
Tue 08 June 2010, 10:45
From the link you posted I went back in the same section, here (http://sensor.baumerelectric.com/productnavigator/scripts/level2.php?pid=&cat=CONFOptoelektronische_Sensoren&ptk=&psg=&language=en&country=US&header=United%2BStates%2B%2F%2BBaumer%2BLtd.&orderby=TM_Opto_Familie|asc&restrict=&specialSort=1) for example model FHDK 10 and many others.

compact housing, small beam diameter, sensing distance adjustable via potentiometer, IP 67

Gerald D
Tue 08 June 2010, 11:26
The model FDHK 10 does not give an output value of the distance. You use the pot to set the "trigger" point . . . . .it behaves like a proximity switch. If the the target is nearer than the set point it could close a switch, and open it again when further away. (It also is not a laser).

One has to select the device very carefully between the many options!

Sergio-k
Thu 10 June 2010, 16:19
What Ries described at his post about laser scanning with a camera,
has been done by this guy : http://www.david-laserscanner.com/

Though the only thing you probably need from his site is the software which is
priced at 230 euros (ouch).

riesvantwisk
Thu 10 June 2010, 16:29
Sergio,

the method I described is not the same, they just use the same tools :)
The difference is that david 3d scanner measures a contour of object and calibrates the angle of the laser with a background board. Knowing the angle it knows more about the objects form based on the contour lines. It's actually a very clever method and easy to do it DIY.

The method described here, and what we talk about really tries to measure the depth. The laser depth measurements calculates the depth by triangulation, it really knows a distance and can do this very accurately, if the laser pointer is fine enough.
BEcause of this we need to have the MM move over X and Y to know the objects surface.

Ries

Sergio-k
Thu 10 June 2010, 17:01
Hi Ries

So to get it straight what you are saying is something like the
depth finder we use at boats for viewing a 2 dimentional sea bottom graph but instead of pulse emition you use laser beams and cameras in a fixed distance ?

riesvantwisk
Thu 10 June 2010, 18:19
Sergio,

a depth finder on boats usually usually uses 'time of flight', on a boat this is done using sound. you emit a pulse from your boat, then wait how long it times it comes back. That time will tell you something about the location of the seabed.

With (single device) laser distance meters that need to be very accurate <1mm, triangulation is used, that's what we discuss here, it also means your measuring distance is short. Time of flight is rather hard for short distances (possible though, just very expensive due to electronics used) time of flight with a accuracy <5mm is hard, you need very fast and noise free electronics.

Laser distance meters for long distances can use time of flight, just don't expect to measure up to or smaller then 1 mm, >5mm, still possible, just electronics are very expensive when accuracy increases.

For us we need something in the sub mm accuracy, our MM can also be that accurate and the objects we can need to be milled that accurate, we also need to take into account any noise during scanning, so we need to have something in the order of +- 0.1mm-0.2mm accurate.

The laser distance meters discussed here can be that accurate, and with the MM moving over X and Y we can scan large objects very accurate. People with a latè can also scan around a object.

Each method has it's advantages, and disadvantages.


David laser scanner is : fixed camera and moving laser to capture contours.
We discuss here : moving camera/CCD, moving laser to capture distance by triangulation.

The advantage of 'our' method is that we can be accurate over large distances. If you have a MM of 2400x1200mm, then over the complete MM you can measure the object at a sub-mm level. If you need to copy ornamentals for restoration, this can be ideal, large areas but still very accurate unto detail, client would be very happy!

David laserscanner loses accuracy when the object size increases, but increases accuracy when the objects get's smaller. This is the reason they can scan complete sofa's, but also coins in very small detail. A other advantage of DL is that they you can scan a object where there are 'shadows' are, making it more suitable for scanning true 3D objects like a head, bust or a shoe ( see gallery (http://www.david-laserscanner.com/?section=Gallery)) .




Ries

Sergio-k
Fri 11 June 2010, 01:19
Ries

First sorry for repeating David's method as i jumped to your post and didn't see that the thread was started with David's laserscanner :D:D
I should start reading from No.1 next time.

I understood about the advantages disadvantages you mentioned above
(or at least i think i did).

So to my conclusion is that we could possibly use DL as a method for
scanning but with a diferent software that would allow us to have the camera
moving at the whole extend of the table and not fixed in a prealligned position.

Something like this : http://www.creaform3d.com/en/handyscan3d/products/default.aspx
but in an affordable DIY budget.

Red_boards
Fri 11 June 2010, 05:11
Please take me back to first principles so I can better follow this thread (it's of interest to me, but let me get a cutting head going first :-).
The scanning device is mounted on the head and then the head is "driven" by software through a xy grid over the object?

Surely if we "drive" the head, then we know where it is, so can reconstruct the height of the head from the object (conversely the Z height of the surface at any xy point)?

More complex: If there is much z variation in the scanned object then this is "eyeballed" and the grid modified to roughly follow xyz needed to keep the scanner within its required distance from the object, but high enough so the head does not hit the object?

Gerald D
Fri 11 June 2010, 06:54
Red, you have it exactly right. The reading head stays at a fixed height, on the y-car, above the bottom of the gantry, which means it cannot crash into the object.

normand blais
Wed 06 October 2010, 17:48
One of those scanner to hook up onto the gantry. 20110-00 South African ? money
http://www.cncsolutions.co.za/laser%20scanner.html

Robert M
Thu 07 October 2010, 10:53
ZAR ( SA curency) to USD $ = +/- 0.145
So = +/- 2910 us$
Robert ;)

Khalid
Wed 13 October 2010, 23:48
Hi,
I have not built the Mechmate yet but i can offer free laser Scanning services to all of the forum members of this great community...
www.free3dscans.blogspot.com
Regards

danilom
Thu 14 October 2010, 01:21
Thanks Khalid!

I have one question, are you using David scanner? Your results are fantastic!

giz
Fri 22 October 2010, 13:43
I've seen Khalid's posts on the ArtCAM and CNCzone forums, and he does do good work. He uses the David scanner.

I wish I could find more information on Streamline Automation's products, including their scanner.

normand blais
Fri 22 October 2010, 16:42
Here is on Streamline automation
http://3dcutting.com/

notimeforwork
Tue 04 November 2014, 09:05
Not sure this would have the resolution you might want but the x box kinect is being used as a 3d scanner. There's lots of info on the Web. I picked one up used for $30 to use - but just have too many other things going on right now to have played with it and report any results...

smreish
Tue 04 November 2014, 11:32
I have used a number of different scan methods for use just as Streamline does.
Actually, modified #5 do cut large statues based on their large machine with removable table and 4th axis.

Streamlines "frog" based series of things uses technology similar to the David principle with a custom interface. Any IR / 2 head scanning system would work for what I see in this discussion thread.

Personally, the hand held david version has always worked well for me for small/med size objects that did not have really super high detail.

2 cents.
Sean