Weitere ähnliche Inhalte Ähnlich wie Automated Low Cost 3D Scanning (20) Kürzlich hochgeladen (20) Automated Low Cost 3D Scanning1. Automated Low Cost 3D Scanning
Nasir Mannan, CCAT
Thomas Tong, 3D3 Solutions
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
2. About CCAT
• CCAT-The Connecticut Center for Advanced Technology, Inc.
• NCAL-National Center for Aerospace Leadership.
• Nonprofit organization that serves as economic development center for
excellence for the region, state and nation.
• Core programs address military and civilian industrial manufacturing
needs.
• Collaborating with 3D3 Solutions to develop push button 3D scanning
solutions in the aerospace and automotive industries.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
3. About 3D3 Solutions
• 3D3 Solutions is a research and development company that develops 3D
scanning technology.
i t h l
• Our focus is on flexible white light 3D scanning systems.
• Works with companies to develop custom scanning solutions for specific
needs such as high resolution, high speed or low cost.
g , g p
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
4. Overview/Introduction
Current Manufacturing Process
• F small and medium sized businesses
For ll d di i db i
• Common metrology process
• Equipment
Issue
• Bottlenecking
Resolution
• I t d ti t non-contact measurement
Introduction to t t t
• Making it easy to use
• Making it cost effective
• H ldi the accuracy
Holding h
• Reporting
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
5. Small and Medium Businesses
• Suppliers to major OEM companies
• Limited on resources and specialized skilled workers
• Most produce parts for OEM’s at high rates
• Must produce parts to quality standard
• Have limited budgets for improvements
• Have limited resources to research new technologies
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
6. Equipment for Quality Control
As parts come off from manufacturing it goes to quality control.
There are two basic pathways for inspection.
Coordinate Measuring Machines Hand Gauging Tools
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
8. Bottlenecking
Bottlenecking occurs when
parts are manufactured
faster than they are being
inspected during the quality
control process.
3 hours
2 weeks
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
9. Bottlenecking
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
10. Resolution: Non-Contact Measurement
Use of 3D scanning to measure parts for quality inspection.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
11. White Light 3D Scanning Systems
• A projector projects a series of
reference patterns onto the scan
object.
• The scene is captured using 1 or 2
cameras.
cameras
• 3D scanning software processes
the images from the cameras to
acquire the data needed to create a
q
3D model.
• Compared to a laser scanner, it
takes a full view of the object
instead of using laser beam.
i d f i l b
• Takes a few seconds to create a
single scan.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
12. Automation
• One button
O b tt scanning i
• 3D scanning generally takes lots of work
(depends on the complexity of the scan object)
• Goal of technology is to save time effort
• Whole object scanning with a single click
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
13. Automating the Pipeline
4. Data
Scan 1. Acquire 2. Scan 3. Scan
Filtering and Report or
Data Alignment Merging
Conversion Manufacturing
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
14. Complete Data Acquisition
• If the object is small and easily moved: Move the object
• If the object is large and difficult to move: Move the scanner
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
15. Multiple Scanners
• The
Th use of multiple scan heads controlled by one PC
f lti l h d t ll d b
• Fast 3D data capture
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
16. Data Alignment
One scanner
• Manual feature alignment
• Motion control alignment
• Photogrammetric dots
Multiple scanners
• Geometric object alignment
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
17. Data Cleanup
Automated Cleanup
• Noise Filters
• H l Filli
Hole Filling
• Scan Merging
• Abnormal Triangles
• S
Smoothing
thi
one scan (left) vs. multiple scans of an object (right)
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
18. Making it Cost Effective: Data Acquisition
Individual
I di id l scans need to be highly
d b hi hl
accurate because downstreaming
cannot improve the quality and
accuracy of the d
f h data.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
19. Making it Cost Effective: Hardware
• Small d
S ll and mediums businesses have a limited resources
di b i h li i d
• Make technology cost effective by using off-the shelf hardware
components (ie. projectors, cameras, lenses)
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
20. Making it Cost Effective: Post Processing
• Range of mesh editing tools to sample, filter, smooth, enhance, decimate,
align, merge, remesh, etc…point clouds and triangulated meshes.
li h i l d d i l d h
• Reverse Engineering modules allows users to create solid or surface models
from scan data.
• Inspection modules allows whole deviation 3D measurement for part
qualification.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
21. Holding the Accuracy: Tracking the Positioner
The coordinate system of the positioner table needs to be accurate in order
for data alignment to be accurate.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
22. Holding the Accuracy: Target Registration
• Align two scans together using features in your scan (ie. spheres).
• Recognizes th spheres and captures th center point of th sphere and
R i the h d t the t i t f the h d
use it for alignment.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
23. Holding the Accuracy: Global Registration
ICP (Iterative Closest P i )
(I i Cl Point):
• Associate points by nearest neighbor criteria.
• Estimate transformation parameters using a
mean square cost function.
• Transform the points using the estimated
parameters.
• Iterate (re-associate the points and so on).
• Provides better accuracy.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
24. Holding the Accuracy: Automation
• Use a positioner table with reference spheres for one button scanning.
• Then align the data using target registration with global alignment.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
25. Case Studies: Accuracy
Single Camera Scanning and Alignment
g g g Simulation Scan - Ideal
46 sec 31 sec
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
26. Case Studies: Accuracy
Accuracy
Single Camera Scanning and Alignment Indicator
scan 1 vs scan 2 scan 2 vs scan 3 scan 3 vs scan 4
Simulation Scan - Ideal
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
27. Causes for Accuracy Distortion
Using a single camera:
Slight rotation creates mis-alignment of
projector and camera coordinate systems.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
28. Solution: Accuracy Improved Using Two Cameras
Improve scan accuracy by using
two industrial cameras.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
29. Case Studies:
Accuracy Improved Using Two Cameras
Accuracy
Two Camera Scanning and Alignment Indicator
scan 1 vs scan 2 scan 2 vs scan 3 scan 3 vs scan 4
Simulation Scan - Ideal
Case study proves that off-the-shelf hardware can produce accurate
results for quality inspection and aerospace at lower cost.
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
31. Automatic Reporting
Acquire Scan
Fixture Part Index Part End of Program Target Align Merge Report
Automate these steps
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
32. Automation Process
10 min
83 hours
h 8.16 hours
8 16 h
8 hour
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
33. Future Plans
• Experiment with different optical techniques/reconstruction algorithms to
E i t ith diff t ti l t h i / t ti l ith t
solve shiny part scanning without coating
• Multi scanner setup for faster quality inspection
• Large area scanning at high accuracies
• Deform mesh alignment – account for movement.
• High accuracy single camera calibration
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.
34. Thank you!
Nasir Mannan Thomas Tong
CCAT 3D3 Solutions
nmannan@ccat.us thomastong@3d3solutions.com
© 2010 Connecticut Center for Advanced Technology Inc. All rights reserved.