A high speed 3D laser scanner for your ShopBot

[tbrown]

Hello all:
I am looking for ShopBot users input and interest level for a 3D scanner for your machine.
In the late 1990's I developed and received patents (#5,910,845 and #5,978,092) for a high speed non-contact laser scanner.
As my money ran short, I had an opportunity to start a small electronics design and contract manufacturing business. My time was consumed by that, and little scanner development took place.
I bought a ShopBot that I used to clip the leads on over a million light emitting diodes for the variable message traffic sign boards that we built. The manufacturing end of the business dried up as my customer move production to Taiwan.
Since then, scanner development picked up as the software has been converted from DOS to Windows, and the hardware has been upgraded.
The "scan head" consists of a video camera, and a pair of laser line generators. The scan head requires precision linear motion that the ShopBot provides quite nicely.
The scanner can be positioned for scanning objects on the ShopBot table, or 90 degrees from the table for scanning human faces (which I have done a lot of) ,or other objects.
The scan head that I use for scanning objects on the ShopBot table, scans a vertical area of about 8 inches. That is, if you attach a video monitor (recommended) to the scanner, you would see an 8 inch vertical area on the screen.
The accuracy of that scanner head is approximately .02 inch in the Y dimension. The X dimension resolution is dependant on how fast the ShopBot moves the scan head while capturing an object image. Moving 1 inch per second yield a .0166 inch X resolution. The Z resolution for the X/Y point is roughly .005 inch. Scan heads for objects such as faces are calibrated for a vertical area of about 18 inches and have a Y resolution of about .045 inch.

The software produces .sbp files or Gcode files. The user can specify the size and resolution of the carving to be produced, as well as the diameter and length of the tool to use.
Several options are available for carving in, out, reversing image, etc., as well as some tool protection options to prevent tool breakage if a large plunge is encountered. There is really too much to try to describe here in this initial forum note. If you have some interest, I can provide more information, or provide a feature you might want.
There is a lot of software still to be completed.
The visual display software needs a lot of work. Software to combine multiple scans needs to be written that will allow larger objects to scanned (larger than the 8 Y inch example).

My target is a product that costs under $3,000.00. That is a lot better than the $20K+ cost of other comparable scanners. Having a ShopBot to provide the linear motion saves a LOT of cost.
Hopefully, there will be some interest out there in the ShopBot world. If not I still have a very cool tool for myself.
Thanks,
Tom Brown

[Brady Watson]

Tom,
I would be interested in seeing actual scanned files and a pic of the object scanned. I would also like to be able to have an editable point cloud and not just a SBP output.

Someone was working on a 3d laser scanner about a year ago that was about $900. It scanned to a point cloud. I haven't heard anything new on it in a while.

-Brady

[billp]

Tom,
How do you do faces with a laser scanner? Do the subjects eyes need to be closed? What is the focal range, or depth of field of focus you expect from your tool?

[stickman]

Tom,

That is the neatest thing I've ever heard of. I'd be interested in seeing some pictures of input, output and the such. Its not in my budget at the moment, but it might be later on.

Jay

[jay]

Tom, I too would be interested in seeing actual scanned files and a pic of the object scanned. I would also like to be able to have an editable point cloud and not just a SBP output.

[daveiannone]

Tom,
Sounds interesting....how easy is it to attach and remove from the Bot?

Dave

[artisan]

I would be MOST interested....D

[pappy]

I am also interested in all of the above points.
I would like to see some of the in/output.

[harold_weber]

Tom, I'm interested, but I certainly want to see some sample objects and scans.

I would also want to have the option of multiple formats for the output data, one format being the x,y,z point coordinates separated by a blank or comma, one point to a line in an ASCII file format.

Not sure of your time frame to have a sample ready, but perhaps you could show it at the 2005 Jamboree in Durham? Also maybe you could bring one to the ShopBot Camps.

[tbrown]

To all,

Thanks for all your input and questions!
Last couple of week have been busy, sorry I haven't responded sooner.

With help from a new friend, I've got the new ShopBot previewer working and will post some images in .sbp format soon. (thanks again Bill)

Have been looking into point cloud output, but have not found a "spec" of the format... Anyone know where I might find it?

Bill P,
The lasers are class II? 1 milli watt.
You should not stare into them, but short exposure to the eyes is not a problem, and your eyes can be open. The lasers project at an angle such that they don't go straight into the eyes.

David M,
I haven't addressed a final mounting method yet. I expect to put a "permanant mount" on the Bot, and have a couple of bolts or pins to remove the scanner head from the mount. I have a PR48, and need to see a PRT before I make final decisions on mounting.

Harold W.
Those outputs seem pretty simple to produce. What do you do with that format?
Jamboree? maybe... ask again in a few months.

Thanks again for your interest.

Tom Brown

[ron brown]

Tom,

I think the X, Y, Z input is a generic script input for many CAD programs.

Ron

[billp]

Tom,
You might try to figure something out that would output into Accutrans (www.micromouse.ca (http://www.micromouse.ca)) They have a VERY slick little piece of shareware ($20...) that takes ALMOST any output, from any device EXCEPT for the Shopbot probe.....They have enough conversion capability to enable most people to have a file they can do something further with, (as in edit!) in programs such as Rhino, 3D Max, etc. I spoke with them awhile back and they were very helpful, and willing to add new features/outputs to their program...

[harold_weber]

Thomas,

I asked for an ASCII file of X,Y,Z points, one to a line because the following programs take that format:

1. Rhinoceros (you can bring the points in and create a point cloud from them)
2. EXCEL
3. Most test file editors
4. SURFER (you bring in the points and can fit surfaces through the points)
5. My CAD program (Design CAD)

With commas separating the coordinates on each line, you have the well-known .csv format

[tbrown]

To all

I was going to post some images today, but found that all the files were to large to upload under the 25 k restriction.....
even zipped...

I have some .sbp files that can be viewed with the version 3 shopbot previewer, some .dxf files that turbocad will make solids from, and .jpg photos of the objects scanned. If anyone would like these samples.... please email me.

I am at the end of a long phone line, with no high speed service available, and would like to email all of you at once. So, I'll wait a couple days to see who responds, and sent to all who respond in one email.

Sorry for the inconvenience.
bdassist@earthlink.net (mailto:bdassist@earthlink.net)

Thanks for the previous file format info. lots of good stuff.
May build some x y z files soon.

Thanks again,

Tom Brown

[Brady Watson]

Thomas,
I'd like to get in on the e-mail.

Thanks!
-Brady

[billp]

Tom,
Please add my name as well..... Thanks..

[stevem]

I'd like to be on the list also.

Thanks

[gerald_d]

Thomas

I will host those big files for you and paste links here, if you want to.

[jay]

Thomas, I would like to be on the email list.
Thanks, Jay Pollard

[stickman]

Thomas, I'd LIke to be on the e-mail list as well.

Thanks, Jay Mack
stickmanwood@hotmail.com (mailto:stickmanwood@hotmail.com)

[fossiera@msn.com]

Tom, please include my name on your email list.

Thanks, Andy

[brett_stahl]

Thomas,
I would like to be on your e-mail list.
Thanks,
Brett

[tbrown]

To All,

Hope you received the files I emailed late last week.
I (and probably you) noticed some anomalies in the tractor roof edge. I have found the source of the problem and corrected it.
The problem that I have coined "bleed through" comes from the arrangement of the camera and lasers. The scanner actually takes 2 pictures of the object as it is scanned. One picture viewed from the left of the object, and the other from the right. This arrangement allows the camera to see features of the object that may reside under other features of the object at the same logical x/z point.
In this case, features of the roof line of the tractor, inside of the tractor cab on the left, "bleed through" to the outside right of the of the tractors roof line.
I thought I had all cases of this covered, but this situation occured in the camera "blooming" compensation, that corrects for how objects appear to get larger as they get closer to the camera.

As I mentioned, the scanner takes two pictures from left and right perspectives. This arrangement captures much more information about the object than is displayed in the output files. For example, you could rotate the tractor image using the scanner software, and produce an output that shows the back end of the tractor as seen by the right perspective. Or rotate the image of me and see the details of my ears (yuck).

Since re-writting the software in windows, I have not re-installed the rotation software, but intend to over the next week or so. I would like to provide some rotated images produced from the same scan files that were used to produce the images you already have.

Thanks to Bill Palumbo who suggested checking out AccuTrans. I have added the ability to produce .xyz files at any specified resolution... feed that to AccuTrans and get cleaner dxfs and other output formats. Still looking into how to produce a point cloud.

I want to provide new corrected files, and new file formats to all who want to be keep up to date.

One member offered to provide these files on his web sight, and I ask if he would, but I have not heard back from him yet. If anyone else can help in this area, please let me know. I hope to get this resolved soon, but have lots on my plate right now.

Thank to all
Tom Brown

[stickman]

Thomas,

Just to let you know, I haven't gotten any e-mail.

Jay
StickMan WoodWorking

[dirk]

Tom, please include my name on your email list.

Thanks Dirk

[tbrown]

Hello all,

Frank has let me place my file on his ftp site.
They can be found at homeimages.net.

Log in as user "shopbot" and password "shopbot"

Frank said he would provide some instruction for those of us who have not used ftp sites... thanks Frank.

There are new image files of the owl, tractor, and me.
They are in xyz format,no commas
.... dxf format
.... sbp format
The dxfs were generated using AccuTrans3D so there should be less problems for you.

The xyz files are ASCII.
The tractor is 142 rows by 205 columns, owl 193 rows by 128 columns, me 175 row by 143 columns.

I also placed 3 .zip files containing only dxf, or sbp, or xyz.
dxfs.zip
sbps.zip
xyzs.zip

I will post some rotated images in a few days.

Thanks again,
Tom

[fleinbach]

The following instructions are for access to my FTP site. Some of you may not need instructions, but there are others out there who do not have very much experience using FTP sites so it is intended for them:

1.

[billp]

Frank,
I have tried your instructions a few times with no luck. I get to the point of the "530 error"and click "OK". But then when I try right clicking or using the "file"command I don't get any option to "login as". What am I missing here? Thanks, Bill P.

[billp]

Frank,
In trying the process again I see that I do not get the phrase "FTP folder error 530 etc." It just says "530 error etc."... Has anyone else had luck accessing these files?

[cnc_works]

Bill, I got the same except when I right clicked in the field after dumping the 530 error, I got the log in dialog, then was able to login and download the files.

Donn

[billp]

Donn,
I think the issue might be that I am using Netscape 7.1 instead of Internet Explorer as my browser..Thanks..

[daveiannone]

Bill
I have Internet Explorer..and it worked just as Frank said

[fleinbach]

I guess I should have mentioned that I use Windows XP. I have not actually tried any other version but I do have 18 freinds who have been accessing my site for the past 3 years and I'm pretty sure some of them are still using older versions of windows. One or two of them did have some problems downloading large files (some download 6 to 700 MB files)but there problem was solved by adding an FTP cliant program called Cute FTP. There are many more ftp programs out there but I have not used any of them. Many of these are free I will look around and let you know what I find.

[brett_stahl]

Some other FTP clients that can be downloaded for free are Ftp Explorer at www.ftpx.com (http://www.ftpx.com) (30 day trial) or Smart Ftp at www.smartftp.com (http://www.smartftp.com). I had no problem using either of them.

[normand]

home made 3d laser scanner http://www.muellerr.ch/engineering/laserscanner/ more in " home made scanner " near the bottom of this link http://www.simple3d.com/ now if I could only be smarter I would not have to wait

[jcrwhiteriver]

Any progress on this lately?

[Brady Watson]

I think that the best value out there right now is the Acculux LR1 @ $1195. You do need to do some work to use it. 1st you need a 12 bit 0-5v data aquisition card (to convert analog to digital) and you need to write some routines to interface with the SB control software, which shouldn't be too difficult.

I spoke to Acculux last week to get a better understanding on how to interface this scanner. 1st thing to understand (or myth to dispell) is that the scanner is going to just sweep over your part with a blanket laser beam and produce results on par with a Cyrax (http://www.duncan-parnell.com/cyra.html). The Acculux in all actuality is a laser ranger...which can only measure distance. No camera, etc. It can sample data up to 1000/sec and it will average it out and then send an analog value to your DA card. In order for this 'scanner' to work, you must set up a routine much like the SB probe, where you define a grid for it to stop and sample distance, write to a file, and then tell the CNC to move to the next point on the grid. It may take up to 3 seconds for the unit to sample distance, average the data and send it to the DA board....then it can move to the next point and do it all over again after the routine has the go ahead from the unit.

Accuracy is not as easily described when speaking in terms of a distance finding laser. Accuracy is affected by the focus of the laser, with the 'sweet spot' being in the center of it's range. In the case of the LR1, this is 2.5" below the LR1's case (range of 0-5" below unit). Accuracy is highest in this 2.5" area. Also, the number of distance samples that you tell the unit to make also plays a part in the overall accuracy. I believe that it is around .01 or so for the LR1. The LR2 has a 0-2" range, boasts a .003" accuracy at 1250 samples/sec. Accuracy is also affected by keeping the scanning range in the 1" sweet spot area.

Theoretically, using an LR1 or LR2, may take as long or longer than the SB probe. Your accuracy is also limited to what your SB can do...You will also find some limitation in what you can scan, as is true with any laser scanner. Ideally the part should be uniform (like gray primer) and show no contrast (woodgrain has contrast). You also can't scan shiny objects since it reflects the laser and gives a false reading.

The Acculux Laser Ranger is a nice unit for measuring distance, but will take a bit of work on your part to use as a laser scanner. I know this has been pursued before (and even implemented) with little feedback posted...so I figured I'd post the skinny of it here.

-Brady

[normand]

another homemade scanner work with rhino http://members.dodo.com.au/~bruzak/scaner/scanneren.htm

[tbrown]

Hello all

I would like to update those who have had interest in my scanner development.
I have been working to make it easier to attach the scanner to my shopbot and insure it is calibrated properly. My first thought was to have the user run some type of test scan, then tweek and adjust... re-scan, tweek and adjust, until the scanner was aligned like it was when initially calibrated. This looked to be a more time consuming, less accurate method than performing some level of calibration. I opted to scan 3 objects and run some calibration software. This takes about 10-15 minutes, then your ready to go. The scans are more accurate also.

I have posted some dxf and jpg files on a friends website that I would like feedback on.

I have reached the conclusion that this scanner may not have the resolution a lot of you are looking for. Object features smaller than the laser beam width are hard to detect. This scanner is better suited to larger objects and was initially developed for faces. I don't know how may of you may have an interest in this type device, but I would like your feedback on the images. In particular, I would like to here from you who sell to the CNC market.

There are a couple problems I am dealing with on a new panasonic camera lens, but expect these can be resolved soon in calibration software.

My electronics business is slow and I am considering taking a part time job to help cover the bills, so development efforts may drop off.

Please visit http://www.moysys.com/tombrown

There is a readme.txt file that has some brief explanition of the files and how the images are created. The dxfs and jpgs are in 2 zip files so as not to use to much of my friends web space.
Select the files and right click on them and "Save As" to download.

I hope to hear from you and will answer any questions you have.

Thanks
Tom Brown
bdassist@earthlink.net (mailto:bdassist@earthlink.net)

[normand]

Hello Thomas great work .How long did it took to scan your face. I dont think you laid on the table so did you fix the scanner on the z and move the x or y while standing in front of it?

[tbrown]

Thanks Normand
You are correct about the mounting. I mounted the scanner on the z, but it could be mounted on the y carriage and moved in the x axis.
The actual scan took about 15 seconds at 2 ips.
On my slow computer (450 MHZ), processing the scan image takes about 90 seconds. About another 2-3 minutes to trim up the image to the desired elevation/width etc, and create the output files.
Thanks again.
Tom.

[wayneo]

Hi Thomas,
Anything new developing in this area lately? I am also very interested in your developments.....
Would like to hear more....

Wayneo

[jay_p]

Are you still around Thomas, and is there any new development in the laser scanner?

Jay

[normand]

another 3d laser scanner pipeline. not as good as a cyrax http://www.rob.cs.tu-bs.de/david/index.html