Digitizing and working a Cloud of Points
MPE asked a question about how to process a Cload of Points from a Digitizer...
This is a very interesting and difficult subject because what you are really talking about is reverse engineering something that is already made. This is common in the racing industry where engine heads and ports are hand ground to make the first port and then folks want to duplicate the other heads to make them all the same.
Its real important that you full understand the actual process and the limitations to the process of digitizing.
Similar Threads:
One of the universities here has a laser type scanner. It is a hand held scanner that uses "GPS" style technology to know where the scanner is positioned. The only problem is it can't scan metal and it can't scan internal. However, I'll take a plug of the port using a pliable silicone(probably from a dental suuply store), and then scan this to obtain the port.
I'll look for the link to their site
http://www.cpai.uq.edu.au/scanning/
Have a look through this
Greig
Now this is an area that I'm reeeeaaaaalllllyyy interested in! I just started doing work for a Pro-Modified Dragster team here in Vancouver Canada. The budget is substantial enough to allow me to get what I need to accomplish our goals.....finally. Anyway, I am looking at a Faro or Romi arm, and accompanying software package to convert the points. I have downloaded an eval copy of Rhino, but I haven't had enough time to learn it yet, though I'm told it's pretty easy as far as solid modelling programs go. I'll keep you posted if your interested.
What kind of racing are you guys involved in? Anyone up for the "One Lap of America"?.
Adam
Fish,
Excalibur has users in all facets of racing...NASCAR, BIKES, JET SKIS, SNOWMOBILES and I'm sure many others I am not aware of.
The key is to get good data...The Faro and Romi arms are not really a good choice since they both use a mechnical probe with a ball tip. This means you are not really getting the point of the surface you are touching but rather the center of the ball on the probe tip.
The better systems will use a laser or optical system that actually returns data points on the surface of the object being scanned.
Not that you can use the mechnical methods, its just they are more difficult and take much more time.
(Note: The opinions expressed in this post are my own and are not necessarily those of CNCzone and its management)
Fish. MPE stands for Motorcycle Performance Engineering. We make components for drag racing really.
The FFR team of Al Miles and Jerry Gordon just won the Canadian(CMDRA) Pro-Drag championship running one of our dataloggers actually.
The datalogger box is a billet aluminium item made using Excalibur.
go to
http://www.stripbike.com/ubb/Forum9/HTML/000610.html
to have a look.
Greig
Here is a device that maybe the breakthrough in laser metrology many of us are looking for:
http://www.aculux.com
Total cost in RS 232 configuration is about $1200. Given, most manufacturers are charging $10000 to include a laser metrology capability, this a real bargain. I am not complaining about their prices... they have to recoup their investment somehow. I just can't afford the price of admission and, fortunately, advances in technology and subsequent reductionsis prices is always on our side.
Check out the downloadable PDF manual for the OEM version. Some basic code and operating data is included in the manual. Any DOS gurus out there willing to develop code to interface this puppie?
I have presented my concept to a luthier friend of mine who has extensive software engineering experience with x86 and pentium class microprocessors who has similar CNC ambitions and he is going to put some time into developing an application for scanning and creating a cloud of points.
Additionally, a web acquaintance of mine tells me he is scanning ing DOS mode at pretty fast rates already in DOS Mode.
I have asked to share his code and will pass along what I can.
Art Fenerty of Mach 1/2 fame is working up support for the serial port in Mach 2 for release in the future. My guess is that if he gets enough interest from the user base he may be inclined to up the priority of development.
I just looked at the Aculux laser site. It is a good idea, but the accuracy is only 0.003". I know the touch probe system is time consuming and a bit of a pain since what you actually get is an offset of the surface, but the accuracy is down to 0.0002" (at least that's what they advertise). I have worked with other conventional CMM's, an this was standard practise. The trick was to have software capable of generating the true offset surface without taking two days to process.
Hey Greig, do you guys have a website?
Adam
Fish....Tell me more. I can live with .003" but I currently have nothing at all in terms of proven technology and will be happy to chuck the model in the machine and let it measure until it's done.
What do you use??
Laser is great for the future. But....a model in the end is just a model.
Fish,
I know the systems say they have 0.0002" tolerance but that is in a perfect world. For the type of work you are doing, I have never seen a Faro do better than 0.005"...especially when you have to then offset the surface which compounds the error.
Perhaps you could get a more accurate version of the Aculux by calling and talking to them. It is always better to work from the original surface data if possible!
(Note: The opinions expressed in this post are my own and are not necessarily those of CNCzone and its management)
All my wooden parts finish with scapers anyway. The last few fractions of a mm are by hand. I just want to be able to prototype in a timely manner and test finished item tolerances.
Guess I need to talk to Aculux.
The Faro and Miniscribes are really too pricey and slow for my taste anyway.
Imagine trying to scan a big "ole" string bass.
Fish we're at www.mperacingonline.com
I'm just curious how you would interface this into a pc to obtain your cloud of points.
Maybe I'm thinking too simplistic here, but here's what I'm thinking:
You would attach it to the spindle in the machine. Run the machine in a series of lines across the job to scan the whole item.
You would need to know the X,Y,and Z value for each scanned point.
So you would to dprint to the PC the x and y location of each movement, and them also somehow obtain the Z value from the laser into the same line of code.
Possibly there is a way to connect the laser in some form af analog or digital form that the controller can use to convert the input into a number and set a parameter with that value so when you do the Dprint, it comes out in the same line of code.
Am I thinking along the right lines here or did I just confuse myself
Greig
Hi guys,
I talked to Acculux yesterday. They said that their products are designed mainly for human bodies. Hmmm........scan the perfect wife.......now there's a project. Anyway, back to earth. They have never designed, and don't have any inclination to design a system with the accuracies that we are looking for.
As far as the conversion to usable info is concerned, Faro and Romi have software that will convert the points (text I believe) to formats that our CAD/CAM stuff can read. I have heard the same thing about the lack of accuracy before, but apparently, the products are much more accurate now. The proof would be in the demo.
As far as, "What do I use now?" someone asked. I use a Mititoyu CMM. It is about ten or so years old, but it is certified twice a year. It outputs points in text files, which are relatively easy to input into various pieces of software as long as you have taken the data from one relative plain, ie; keep one axis constant. I have never had any real difficulty creating a pretty accurate model from these points. Well, really, the dupicate is a result of all the errors combined, machining conditions included.
The reason we want to change from a standard CMM is that it is too difficult to move some of the parts - they're simply too big and heavy. I have a Renishaw in the machine, which I use to verify measuremnts for the first go 'round of measurements, but not good practice to rely on only one measuring system, I need to have verification of my work. Or rather, my customers need to...
If we can put a man on the moon, than why can't we get a simple....you know what I mean. (Did we reeeaaallllly put a man on the moon?) I smell a conspiracy......
I know I am all over the place here, but, the coffee has run out. As far as needing to know some value before scanning with the laser system is concerned, I think it wouldn't really matter. All the points would be relative to each other. The cloud can be opened in your CAD program, and moved to whatever datum you decide. That's how I do it now. The CMM starts from one spot, but it doesn't matter where that point is. As long as the part is stationary, the points would just be increments from eachother.
Adam
Quote "They said that their products are designed mainly for human bodies. Hmmm"
What a bunch of marroons? They don't seem to be pushing too hard to make a sale!!
They clearly haven't read their own publications very well. The user manual for the RS232 model clearly targets manufacturing processes.
Additionally: They claim a resolution capability of .001" and accuracy of .01% (1 Standard Deviation) at a 500 hz rate. That's pretty good in my book (well within my requirements).
I'll continue my research and share my successes (and failures). Maybe we can get something good going here.
Hey twombo,
You're right, they don't seem too interested in making a sale, or at a new market for their products. The guy that called me just seemed to want to get off the phone as soon as possible. Maybe we could take their idea, and make a real product out of it!!
I don't know if I have this right, but, resolution and accuracy are different beasts. Resolution of 0.001 means that it can recognize features or differences as small as 0.001. But accuracy requires that the system will be able to place the features in relation to eachother, or a datum, within a meaure of accuracy that they are claiming. I look at it this way, I can machine any feature to within a couple of tenths as far as the readout on my machine is concerned, but it my machine accurate enough to claim that it is actually within a couple of tenths?
On that note, the industry has almost gotten to the point where the accuracies requested are barely measurable, let alone repeatable. I have been in shops that claim micron accuracies, but I think they're full of crap. Unless the part is measured on a certified, high quality meausuring instrument, (like a Zeiss CMM), in a temperature controlled atmosphere, the claim is just hot air. I'll get off my soapbox now......
Let me know what else you find, and I'll do the same.
Adam
Hey Fish, I'll have you know that every part I produce is exactly the size that it is. You cannot even detect a discrepancy between what it is and what it is.
First you get good, then you get fast. Then grouchiness sets in.
(Note: The opinions expressed in this post are my own and are not necessarily those of CNCzone and its management)
Does anyone have any more info on machineing from point cloud data , like how do you get from a .stl format to something I could use like maybe an igs or something.
VIPERDM100
Viper,
STL is not really a cloud of points... Sometimes software canconver cloud of point data into a STL. STL can still be worked with if you don't mind working with a larger amount of really small surfaces.
Excalibur will let you import the STL into a CAD model. CAM can then process the data for some operations...most surface type machining.
(Note: The opinions expressed in this post are my own and are not necessarily those of CNCzone and its management)
Posting Permissions
Reply with Quote
