RE: 3D Sensor for Reverse Eng

From: Derek_Smith-EDS014@email.mot.com
Date: Fri May 17 1996 - 00:15:58 EEST


Hong,
     With regard to accuracy, keep in mind that this is a statistical value.
     Consider a "known" article for which you have XYZ coordinates for 100,
000 points for one square inch of a surface. Now, you scan the same area
with your scanner, taking 100,000 points which correlate to your "real"
values.
     If you now subtract the measured or "scanned" point values from each
"known" value, you will obtain a list of deviations from actual.
Statistically this data should form a normal binomial distribution.
     Ok, so as to the accuracy of the system. Is the accuracy going to be
the worst case, the average, or described in terms of standard deviations
from the measured distribution. The standard in the industry appears to be
in terms of standard deviation, yet there is a BIG difference here as well.
Some vendors publish a machine accuracy value which is 1 SD, others 3 SD.
In my opinion, 3 standard deviations, or 3 Sigma, is an acceptable form,
and describes a range from actual which includes 97% of your data points.
     Another aspect to consider is the shift in the measured mean from
"actual". I did not acquire any of this data in my investigations. The
most accurate system I found was the Hymark system, which was also one of
the fastest and easiest to use, but unfortunately, way too pricey based on
your listed cost criteria. It really depends on your applicaton. I was
investigating systems which would be good for "hand-held" sized parts which
have been injection molded.

E. Derek Smith
Motorola Radio Products Group
Plantation, Florida
eds014@email.mot.com

________________________________________________________
To: kaithong@tp.ac.sg@INTERNET
Cc: rp-ml@bart.lpt.fi@INTERNET
From: hding@alnitak.usc.edu@INTERNET on Thu, May 16, 1996 4:37 PM
Subject: 3D Sensor for Reverse Eng

X-Authentication-Warning: bart.lpt.fi: major set sender to owner-rp-ml using
-f
Organization: IMPACT Laboratory - University of Southern California
X-Mailer: Mozilla 2.01Gold (WinNT; I)
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Precedence: bulk
X-Mailing-List: rp-ml@ltk.hut.fi

Dear Friends,

We are very happy to receive so much advice from
friends (Wohlers, Mitchell, Bauer, Daniel,
Addleman, Mariner, just name a few) since we posted
our message "3D sensor for RE". We would like to
take this opportunity to thank all of you. All the
advice from you are really helpful to our search of
3D sensor products.

Some of you have mentioned the experience with
Digibotics scanners, we would deeply appreciate it
if you could share with us following first-hand
data in the use of Digibotics (particularly their
model II) from a client point of view:

1) It comes with a editor for you to manipulate
with the scanned data. Does the editor only serve
the purpose for fixing "bad points" in point
clouds ?

2) We hear from someone that the claimed accuracy
is +/- 0.001 in, not 0.01 in, is that true ? what
about the real accuracy the system normally
achieves ?

3) It can output STL file. Some of you have
fabricated the model on a RP machine. How about the
quality of the STL file ?

4) It can also output IGES file. Has anybody tried
to manipulate its IGES data using a CAD system,
like Pro/E ? How much space there for you to
modify this IGES data ? How much have Digibotics
guys done with "point clouds -> STL and/or IGES" ?
(BTW, what version of IGES does it support ?), The
idea case for us is that we want be able to
manipulate its IGES data (to some extent) using a
CAD system to produce a revised model which is
different to that extent from the original scanned
object. Is that achievable by using the Digibotics
II or any others ?

5) We don't know whether it is appropriate for us
to raise the price question here. What is the
normal price range for a Digibotics II ?

Thanks again.

Hong



This archive was generated by hypermail 2.1.2 : Tue Jun 05 2001 - 22:37:20 EEST