Dirt Talk: More Straight Talk about TIN Surfaces & Machine Control
by Bill Crumb
April 8, 2010
I’ve been saying for years we need new algorithms for 3D
model creation, so in this blog I want to continue to open up this discussion.
A 3D TIN model is a triangulated irregular network, which
means a network of triangles that connect together to make a “best fit”
surface. These TINs are created by model-making professionals, and every model
maker out there knows there must be a better way to smooth out the triangle
junctions. Unfortunately, in most cases the model maker is controlled by the
data that the engineering firm sends him. And the model maker may also be
restricted by the software he or she is using to build the model.
As in every profession, the degree of professionalism of
each model maker varies from person to person. But to achieve the highest
degree of professionalism, a model maker must be able to consider a site from
multiple points of view--namely, that of the engineers, surveyors, grade
setters, grade checkers, inspectors and operators.
In my opinion, our current efforts are like trying to make a
disco ball perfectly round. We are trying to implement human desire with
mechanical reality, and triangles are currently our best fit solution. I wish
we could figure out how to dip that old disco ball in a liquid solution and
develop a smoother model. The real issue, though, is that the automation
manufactures are trying to best fit a dynamic machine control system to a
best-fit site model. We could tighten up the triangles, which generates a
smoother surface. However, by doing so the file size becomes a problem. We then
run into control boxes and data processors that cannot perform fast enough,
which leads to system latency. Imagine cutting grade in second gear, abruptly
stopping and the cutting edge raising 0.12’ off the ground…have you been in
that situation and then questioned which on-grade light was correct, the one
down there or the one up here?
I have experienced a 3D model that was supposed to be a 2%
crowned road change (while in motion on the machine) from 2% to 1.7% and back
to 2% in twenty feet. This may not sound like much, but in a 10’ wide cutting
path that’s 0.03’, or daylight under a cutting edge every twenty feet after I
turned around. Now here’s where this subject truly gets interesting. The
contractor will call the local automation dealer and request his on-site
presence ASAP! Then one of two things usually happens:
(1) Parts get swapped, an invoice is generated and the
problem seems to get better. (But it’s worth noting that the operator is
usually in a different place on the site.)
OR
(2) The automation technician points to the model and says,
“Not my problem.” And while the head-scratching and finger-pointing is going
on, the site production is on hold, which results in a frustrated customer.
So, to sum it all up: the modeling software developers are
currently trying to best-fit a usable model with a small data file, while the
automation manufactures are trying to best-fit their machine control systems to
the site model. By doing this, our current solutions become larger site
tolerances which result in more finger-pointing possibilities!
Could this be avoided if we changed our point-to-point
mentality?
A final note: AMW (Alan
& Mark Williams) is the software foundation for Advanced Geo Positioning
Solutions Inc.--better known as AGPS. The AMW software uses vertical and
horizontal curve technology and is currently the only software I know of that
truly builds an alignment/stations and cross section model. This software uses
true cross sections attached to stations and--and they remain cross sections
once the file is built. Yes, many 3D software companies say they use an
alignment and cross sections, but the end result is a TIN. This is not the case
with the AMW software.
Got any feedback?
Bill Crumb
Geotrainer@aol.com
Bill Crumb has been involved in the machine control industry for more than 35 years. He first started operating a Cat D6D with a laser control system in1978. Since then he has worked for a number of manufacturers and currently owns and operates his own training business, Machine Control Technology Inc. (www.machinecontroltechnologyinc.com) based in Meeker, Oklahoma.
Comments (3)Post a Comment
Did you enjoy this article? Click here to subscribe to the magazine.
Title: Parametric Roads
By: simon
Posted: April 12, 2010 5:38 PM
It's true that most systems do not use a truly "parametric" or template-based road model, meaning one that is made up of a horizontal alignment, vertical alignment, and templates. Trimble's solution can use a truly parametric road design and has for years. It can also use a surface-based road design as well. It should be noted that if the surface is properly densified, using a template-based or surface-based road model is irrelevant.
Title: Parametric Roads
By: simon
Posted: April 12, 2010 6:29 PM
It's true that most systems do not use a truly "parametric" or template-based road model, meaning one that is made up of a horizontal alignment, vertical alignment, and templates. Trimble's solution can use a truly parametric road design and has for years. It can also use a surface-based road design as well. It should be noted that if the surface is properly densified, using a template-based or surface-based road model is irrelevant.
Title: I agree
By: Bill Crumb
Posted: May 9, 2010 11:29 AM
Simon, Thanks for the imput... I've looked into Trimble's solution and agree... and will also say, The "Trimble Business Center" also impressed me"