Threaded View
Subject
David McMillan posted on September 13, 2008, 8:44 am
This is a bit of a long shot, but so far all of my regular suppliers
have been coming up empty.
I've got several catalogs offering small self-contained laser sensors
that "stripe" a single laser line across a target surface, and generate
a map of the height of the surface where the line reflected. So, if you
targeted this laser line across a countersunk hole, you could get a
perfect cross-section map of the hole, and precise measurement of the
hole diameter, countersink depth and slope, etc.
But, of course, this only works if you can put the laser line across
the diameter of the hole. If you're off-center, with a single laser
line, you'll measure what appears to be a smaller hole, and never know it.
What I need, and haven't been able to find, is a unit like this that
*also* scans multiple lines, or sweeps the single line across a target
area. Basically, I need to measure the hole, but I also have to *find*
the hole -- my tooling situation prevents me from having the hole placed
in front of the sensor in a repeatable fashion.
I *could* duplicate the effect I need by mounting one of the
single-line scanners on a panning servo and sweeping it across my target
area, but that gets kludgey and unreliable, not to mention bulky -- I'm
volume limited. I can't believe that no one makes a self-contained
sensor like this, but so far no one seems to have anything on the
market. Heck, they do it already with supermarket bar-code scanners,
why not industrial surface-mapping sensors?
My requirements are pretty modest -- maybe 120-150mm range, and a
scanning area of around 10mm x 10mm. The hardest requirement is that I
need accuracy of around 0.1mm.
Anyone recall seeing any likely candidates they could point me toward?
cadcoke4 posted on
Re: Scanning laser
While I am not totally clear on all your issues, here is a desktop
laser scanner that seem to emply multiple lasers, and therefore migh
do a better job. I am unclear if also employs mutliple cameras, which
would seem to be necessary to catch all the geometry of a hole. But,
perhaps a position change of the sensor or object can fix that issue.
https://www.nextengine.com
It is about the size of a shoe box, which may be too large for your
application. I haven't seen the internals. But, perhaps the device
can have it components removed and adapted to your situation.
Joe Dunfee
Re: Scanning laser
cadcoke4 wrote:
Ooooo... I *want* one. But it's waaay overkill for my application, in
size *and* capability.
Hmmm, let me see... okay, here's an example of the kind of
one-dimensional line scanner I mentioned:
http://www.micro-epsilon.com/en/Measurement-devices/profile-sensor/
It returns a high-precision height measurement of each pixel of the
line drawn on the surface. I actually experimented with mounting this
sensor on an actuator to "pan" the stripe and generate what amounted to
a topographical map of a 2-D area, but we would have had to post-process
out all the resultant data ourselves (plus, the sensor was, again, just
too darn big), and we were really looking for a turnkey solution.
What I need now is less of the topographical mapping capability, and
more the ability to locate the center of a hole that may be of a random
size, appearing at a random location within my hypothetical sensor's
field of view.
Gordon McComb posted on
Re: Scanning laser
David McMillan wrote:
I would think that it's not the scanning mechanics that are the main
sticking point here, but the imager. Certainly these imagers exist, but
writing the software to read back the values is a full time job in
inself. So, I think a ready-made solution is the preferred method if you
don't have a lot of R&D time/money available.
If you do take a line scanner and create a laser raster scanner I'd use
a high-quality galvanometer (especially one with a high resolution
absolute encoder feedback), and not a servo motor. Galvos are much
faster, more reliable, and more accurate for this kind of application. I
guess if you're going to use one galvo you might as well use two, and do
away with the spinning mirror common in line scanners. You control the
galvos with a pair of sine waves, and you have absolute control over the
beam. Ready-made devices do exist here (laser light shows), but these
might be more bulky than you'd like. And, you're still lacking the
receiving sensor.
-- Gordon
John Nagle posted on
Re: Scanning laser
Gordon McComb wrote:
We considered that once, and it's still a good idea, especially if
you don't have to deal with the problems of trying to get 50 meter
range and operating on a moving vehicle. Also consider MEMS
scanning mirrors, which are a related technology in IC size.
http://www.ipms.fraunhofer.de/common/products/MSD/msm-e.pdf
http://www.ipms.fraunhofer.de/common/products/SAS/Systeme/3d-metrology-e.pdf
The problem with building a LIDAR is that it's a hard enough job
to be too much for a one-off. You need electronics skills, software
skills, machine shop skills, optical assembly skills, and a rather
broad range of tools. The reason scanning LIDAR units cost so much
is that the engineering costs aren't spread over many units. They're
not really that complex; a cell phone is far more complicated.
John Nagle
Re: Scanning laser
Then you don't need to map the entire hole. Simply getting a part of
a hole should give you enough information to locate its center. Any
one-laser one-camera type of scanner should be able to provide a
sufficient map to extract this information.
Perhaps you can make certain assumptions that will help you. I.e. the
surface is flat and you know its orientation. With this assumption,
and the raw point cloud of your scan, the edges of the holes will all
be at points that have multiple points lined up in the Z-axis. Then,
other points on that edge will be adjacent to the first multiple-point
location. After you've gathered all of the points on the arc, then it
should be simple to extract the center.
Joe Dunfee
Curt Welch posted on
Re: Scanning laser
Maybe, you should drop all the fancy laser scanning and complex software
and just locate the hole by touch?
--
Curt Welch http://CurtWelch.Com/
curt@kcwc.com http://NewsReader.Com/
jpizano posted on
Re: Scanning laser
Have you looked at the product offering from Sick?
http://www.sickivp.se/sickivp/products/en.html
Re: Scanning laser
David McMillan wrote:
Back when we were doing a DARPA Grand Challenge vehicle, we looked
into having something like that made, but it wasn't worth it as a
one-off item. We'd considered a rotating-prism scanner with each
face of the prism at a slightly different tilt. Here's some
of the design data from back when we were considering this.
http://www.overbot.com/grandchallenge/note21.html
http://www.overbot.com/grandchallenge/note22.html
We needed a big prism because, for long range, we needed big
collecting optics. For the application here, you don't need
that, and could use a much smaller prism.
This stuff is 6 years old. There's been some progress since then,
and more devices are available. But small, low-cost, and multi-beam
is still hard to find.
John Nagle
Re: Scanning laser
John Nagle wrote:
Oooo, FUN stuff!
It's definitely an idea. I'm *really* looking for a turnkey solution
(this job has enough R&D already), but needs must, etc.
Yeah. ISRA makes a unit that would almost work, but it's huge, and has
lots of additional functions that would go to waste for this application.
Daniel Rudy posted on
Re: Scanning laser
At about the time of 9/13/2008 5:44 AM, David McMillan stated the following:
I have personally given this some serious thought for along time now.
The way the supermarket barcode readers work is that they use rotating
optics to scan the beam across the field of view. You could probably do
something with a rotating mirror of sufficient length along the
rotational axis. The two issues that I see with this is the flood of
data would require some serious processing power to do this in
real-time. The second issue would be keeping the data straight since
the mirror is constantly moving.
--
Daniel Rudy
Email address has been base64 encoded to reduce spam
Decode email address using b64decode or uudecode -m
Site Timeline
- » AI Mind Exhibit
- — Next thread in » General Robotics Forum
-
- » Positions and frames of reference
- — Previous thread in » General Robotics Forum
-
- » evoMUSART 2013: First CFP (with correct dates)
- — Newest thread in » General Robotics Forum
-
- » Heat pump refrigerant change to R-22 substitute
- — The site's Newest Thread. Posted in » General Metalworking
-
- » DCC sound question
- — The site's Last Updated Thread. Posted in » Model Railroad Forum
-
XML Sitemap