laser digitizing

[owhite]

Originally posted by twombo Who out there .... I want to implement this capability.... anybody out here (in here?) got any experience with this little jewel. http://www.aculux.com/ Like I said "if it looks too good to be true" just what is it??? twombo

I contacted the aculux people and they want $2,495 for their device.

Owen

[GWheeler]

The Disto Plus is apparently $699.00 list, but getting the data out of it through Bluetooth might be a chore. Each data point would takes .5 to about 4 seconds for the machine to aquire. Getting a "point cloud" of a large data set would take a bit of time.

[cadcoke3]

I use a Microscribe digitizing arm at my work. It is definitely the lowest cost solution of its type out there. We purchased our on E-bay for around $800, which is a very good price. I think a more average price used is around $1,000.

However, I wouldn't say that the touch probe is the easiest way to digitize stuff. If want to digitize corners, it is difficult to put the probe point on a fine corner (I ended up machining a special point with a "V" to help). Think of the probe arm as a 3-D mouse.

I'm sure laser or even the gantry-mill type touch probes are better for copying an existing object in many situations. The challenge is that they produce too much data, and it takes expensive software to really examine the point clouds and turn it into simplified geometry.

Joe Dunfee

[CNC Pro]

How often does a home shop require an object laser digitized? How much detail? Does the model need to be scanned in multiple scans, and then have those scans pieced together? Would outsourcing the job to a firm with the hardware & software be an option?

[cadcoke3]

I don't have a home CNC machine, so my answers are not from the perspective you are looking for. Rather, I work at a theater where I design the structural aspects of the sets. My job starts when they hand me a 1/2"=1' scale model of the set piece they want. So, my purpose is to get that model into my CAD system.

I think I have posted this here before, but there are several home-brew laser scanner projects out there on the net. However, the missing aspect is the ability to put multiple scans together. It seems laser scanners have an effective coverage of about 90 degrees of a curved object. I.e. a cup would only effectively get 1/4 of its circumference scanned.

I suspect most home users would need to combine two scans, at the least. But if the approach is to create a very large point cloud, and then try to organize or simplify that data... that seems to be too challenging even for the most expensive software. Those that make the home scanners simply things considerably by analyzing only one line of the scan at a time. Then their task is only to extract the curve for that line.

I think the challenge is to try to figgure out a way to do that same easy curve extraction with multiple scans. I think the way to do that is to make sure the laser scan plane is kept on the same plane for each scan.

If this were a device to put on a CNC machine head, it might be accomplished by using two lasers, aimed at the center of the scan and angled from each other by 90 degrees. This would provide coverage for 180 degrees around a cylindrical shaped object. Then to see that line, also provide two separate cameras (USB cameras are pretty cheap nowdays) to see that same scan line from different angles.

This is not a method that is guaranteed to scan every nook of every shape. But it should be reasonably easy to implement without a lot of intense processing of the data.

[NewCents]

Found this, thought it was pretty cool:
http://www.indoor.flyer.co.uk/probe.htm

[cadcoke3]

That web page was very nicely done. I've seen touch probes before, and they are certainly a nice addition to a milling machine. If the touch probe has an end the same diameter as the cutting head, it is possible to use the tool path taken from a direct copy of the digitizing path. The shopbot has a low-cost probe available.

[GWheeler]

Found something at the Mark III Robot Store that might work for CNC, and is cheap.

http://www.junun.org/MarkIII/Info.jsp?item=30

"Sharp GP2D02 Distance Measuring Sensor , $12.50
Infrared distance measuring sensor. Accurately determines range to target between 10cm [ approx. 3"] and 80cm [about 2'-8"]. Output is an 8-bit binary number representing the distance, read serially. Comes with one 1N4148 protection diode. "

It is used in cameras and stuff to measure distance. If the object isn't shiny or too reflective or irridescent (it confuses the sensor) it might be very handy for inputing data. Am not sure how accurate it is, but it might be able to scan in Z heights quickly. Without having to touch the object.

The infrared light would have to be shielded in some kind of flat black light absorbing tube to focus the light to a small point on the scanned object. Or perhaps a infrared laser could be used?

[Graham S]

I don't think it would be overly good for digitizing as there won't be much localization in x and y axes. Interesting item though and might well be handy as a sort of adaptable z limit switch perhaps?

[GWheeler]

Am not sure how you define localization.

I'm thinking that with the proper tube it might project a beam of about an eighth of an inch in diameter. Part of the sensor uses an intregal light sensor that works as a camera to triangulate the angle to the point of light and estimate distance. The offset distance from light projector to light sensor is a fraction of an inch, so distances less than a foot might be accurate enough. This would probably have to be done in a darkened room, and the object might have to be painted for the best results.

The machine would have to do some kind of grid search pattern. The beam would hopefully be mounted perfectly perpendicular to the x and y axis. But, if not, it might not matter much. Any small angle off-perpendicular would be consistant throughout the search pattern, so it probably wouldn't be a problem. It would be the same as a slight offset in the true object location.

Anyway, it looks like it would be a lot cheaper than a laser setup. And wouldn't require laser safety glasses either.

[Graham S]

"flat black light absorbing tube "

I didn't understand this to be honest unless you mean to try and remove all but the most collimated rays with a tube.

The sensors do not respond only to the central point being scanned. If you try and make the light into a beam then if you imagine it comming down at a single angle and then lift the sample (same thing as scanning two points of different height) the beam will traverse the sample and probably move off the sensor. If you make the source vertical then you get no light back to the sensor. Also if the object is angled it will divert the beam and give a false reading.

A laser setup (and do we know exactly how they normally work for digitizing?) might not be expensive if you base it on a laser pointer. You don't need goggles just a bit of card to put around the digitizing process, if that.

Graham

[HenryCNC]

I'm too lazy to make a drawing right now, but I think the way most laser probes work is to point the laser straight down from the spindle. Then a camera (one or more) look at the dot from an angle. The software finds the center of the projected spot - the position of the spot is used to calculate the distance from the spindle.

There are a couple issues to work around:

1) laser light causes speckle. The light interferes with itself and causes that "grainy" look. Normally, I find that kind of cool, but the software will have to deal with this. Adding a filter on the camera so that only the laser light reaches the sensor helps.

2) The bigger the angle between the camera and the laser, the more accurate the system will be (up to 45 degrees). However, this makes it difficult to measure into recessed areas. Trial and error would have to determine the best compromise.

3) Historically, laser probes have a difficult time dealing with edges. As the laser aproaches the edge the dot gets smaller and smaller. At the same time, the part of the beam that is falling over the edge has to do something... and that causes trouble.


I'm very new to this whole CNC world. Is there anyone currently working on a DIY probe? Maybe I could offer some help...


Henry