on code.

This article was written for http://fablab.waag.org/

1. Introduction

At the Waag Society, and Fablab Amsterdam, together with Miguel Jimenez andKarim Amhali I have developed a 3d laser triangulation scanner.

A short introduction into 3d scanning can be found here
, also, thefollowing presentation
contains lots of useful information. The wholesetup is built around a webcam, a rotation platform and a focussed line laser.By taking pictures of the object from each angle while project the laser atthe object under a know degree, we can determine the position of points inspace by measuring the distance of the laser line from the center of theprojection.In our final design of the scanner, we decided to add a number ofimprovements, such as an arduino controlled rotary platform and a second laserprojecting at the opposite angle, to capture more detail and fill possiblegaps caused by self-casting shadows from the object.You can find software and some more resources at the follow url: https://github.com/mvhenten/pylatscan
tscan&sa=D&sntz=1&usg=AFQjCNEw4Kp_SM-u147iTIWFZ1HA3rGmGAThis document describes in somewhat more detail the construction of a body forthe 3d scanner out of plywood.

2. Considerations and partlist

The current hardware design for our prototype has evolved from extensivetesting with earlier setups. A laser angle of about 30 degrees seems to be themost optimal position for the laser, and moving the laser to different anglesautomatically might improve scan quality, but increases the complexity of theoveral design. Using a single laser proved to be insufficient, as objectsprotubing too much in the horizontal plane tend to obscure the laser lineunder certain circumstances. Therefore, a secondary laser was added under theopposite angle of -30 degrees.Fitting the lasers rigidly as to stay in place during transport is vital forthe usability of the scanner, since calibrating the whole setup is crucial inorder to produce good results, but can be quite cumbersome.The lense of our camera must be centered at the exact center of the rotation,but moving the camera up and own or closer to the object is needed to scandifferent sized objects, and to move the focus point of the camera to thevertical center of each object as much as possible ( to produce the bestresults, an almost orhogonal projection is the best kind of input.)Earlier designs used plexiglass, this current design, however, uses plywood asthe base material. Plywood is nice to work with, since you can cut, drill andsand it into shape very easily, using both the shopbot and more conventionaltools such as a chisel and a hammer. This means the overall material is moreflexible and you can work around small mistakes or omissions in the design.

2.1 Platform construction – partlist

Some of the bolts and screws could be exchanged for other parts you may havelying around – it just happened to be the pieces we could get our hands on.The SVG files for the shopbot also contain holes for the bolts, so keep thatin mind when taking different sizes.The M6 pins may not have been ideal in this case, since a thicker guiding mayprovide more stability.

partlist:

1. 19mm plywood. dimensions of the platform are around 40x50mm and you needtwo sheets
2. “Standard” 40mm “flightcase” or “tool case” suitcase
3. about 50mm of 10M steel bar
4. about 25mm of 6M screw wire
5. one long 6M bold ( I found 12 cm the longest I managed to find )
6. two long 6M pins ( The kind with a cap on the end )
7. “draai tappen” ( or wood pivot, the kind you can slam into a piece ofwood )
8. various bolts, nuts in 10M/6M
9. 2mm round- head wood screws

2.2 Scanner hardware – partlist

Unfortunately the part number of the original stepper used in our prototypesgot lost, but I’ll assume the farnel part will do just fine. Also, someexact details ( such as the type of white leds ) got lost as they were notpicked by me. The led mount is made of a plexiglass scrap and hot glue. At thecurrent time, I don’t have any connection schematics but I will provide themas soon as I have constructed the final prototype.

1. 2 Green line lasers with focus ring ( LC532-5-3-F(16×65) available from laserfuchs.de )
2. 1 Logitech Webcam Pro 9000 ( The better, the best!)
3. 1 Arduino board
4. 1 Arduino motorshield
5. 1 Stepper engine, 400 steps (for example ST2818S1006-B from Farnell)
6. 6 bright white leds
7. wires to connect everything

2.3 Schematics, blueprints

Blueprint files are available from github, use the .eps for the import in Illustrator.

The shopbot cutter we used was a 6 mm, which is pretty precise. The designdoes not include any depth-milling, but smaller is better, since more precise. The 6mm cutter just happened to be on the machine when I used it.

3. Construction

To explore the capabilities of the shopbot, I designed a “snap fit” lasermount and platform fix that could be cut out of one piece of plywood and put together as an IKEA package. The platform consists of three layers: top, middle and bottom. The middlelayer can be used to guide cables and provides extra stabiltiy to the lasermounts, the bottom layer is for finishing and fixing the electronics. Overalldimensions were choosen to be able to fit into a standard size suitcase, in our case the biggest affordable aluminum suitcase we could buy at a localhardware store.

However, to be honest, each layer was designed along the wayin an iterative process.

3.1 The Mark One

Once the cutting was done, I constructed a camera mount out of plywood andsome large bolts that could be moved up and down by turning a screw. A strongmotorcycle spring provides some rigidity for t his construction, and it provedto be rather stable. I used standard bolts so I could fix the bolt headseasily in a cavity on the bottom sinde of the the second layer. All this wasdone by hand, improvizing along the way, but most of this work found it’sway into the second attempt. Remember to keep a small chisel at hand for thistype of work!  Plywood is excellent to work with in this way

 

Once this was finished, I started to fix the lasers on their mounts, this iswhere I realised something had gone wrong in the milling – the cavities for the lasers were too large. I finally found out that this was caused by the transition from inkscape SVG to illustrator, for some strange reason. This canbe avoided by us

.

I improvised again, by drilling 12mm holes into the other side of the mounts.( it is very important to do a precise job here, since the laser angles mustbe know and exact! ). I calibrated and ensured that the lasers were at astraight 30 degree angle by switching them on. The two lines should merge atexactly the center of your rotary platform! Again, some of this was improvisedin the first attempt, for example by adding a layer of black ducktape aroundthe laser casing to allow for more grip.
After this, it was time to fix the camera onto the mount. The packaging of the9000 pro is not the most ideal for this setup, as it is hard to center thelens and keep the whole thing straight and horizontal. One important thing todo here is to switch on the lasers and camera, and keep looking at thepicture. I’ve written a small python script that shows a line trough thecenter of the image, and while reassuring that everything stays straight andcentered by looking at the screen, I fixed both camera’s, laser and platformin one go.

3.2 Lessons learned: the Mark II

From the inital built, I learned that:

1. The shopbot is much more precise then I had anticipated. don’t hand-
   drill. trust the shopbot!
2. The camera mount works pretty reliable for up-and-down movement. However,moving along the Z axis is a very nice to have, since it’ll allow forscanning small objects.
3. The snap-fit approach works pretty nice, and plywood is sturdy anough forthis purpose

Altough the “Mark One” can be used within the project I intended it for,I’ve started work on an improved design ( this time without the scaling bug) that will actually fit the suitcase as intended.I started out by making a constructive test of a Z-Y axis mount, much like aCNC milling machine provides.

This mount eventually made it into the next design, the top layer already cut.[][24] [][26][24]: http://www.flickr.com/photos/57913158@N05/5329971272/ (Assembling theZY axis system – it’s a fit! by mvhenten, on Flickr)[26]: http://www.flickr.com/photos/57913158@N05/5329357199/ (3d scannerassembly by mvhenten, on Flickr)…more to follow!