2. Company Overview
• Founded 2005 in Irvine, CA, USA
• Training & Solution Center
• Metrology Consulting
• Universal Metrology Automation®
(UMA)
• Streamline Software
• Spotlight AR++ Software
• Part ID Software
15. • Manages Device
Connections
• Manage Part List
• Traceability
• Robot Teaching vs.
Programming
• Scanner settings
• Data alignment and analysis
• Reporting
• User Access Control
• Touch Screen Interface
(HMI)
16. • SCAN-to-CAD Comparison
• >30,000 LicensesWorldwide
• Standardized by industry
leaders
• Smart GD&T
• Parametric Reporting
• Multi-piece inspection
• Statistical process control
Welcome
Introduce
Purpose of this webinar is to expose you to a wide variety of applications of automated inspection.
Different types of inspection – Dimensional, cosmetic, assembly verification
Dimensional inspection will focus mainly on 3d scanning
Defect detection and assembly verification will be focused more on 2D machine vision.
Before we jump into that I want to give a brief history of the company so you can see how we started and how we got to where we are now.
Started in Irvine in 2005 as the master distributor of polyworks.
Metrology training and consulting.
Handheld 3d scanning was easily accessible, but customers kept asking for automated scanning.
We worked on some custom projects, but as you know, custom projects are expensive and take a long time.
We wanted to create some standardized solutions that can be used by the general market.
So we created our Universal Metrology automation platform (UMA) which allows us to use any scanner, any robot, other motion controllers to quickly deploy automated scanning solutions. Our software, Streamline, is what allows us to do this.
Over the years, we’ve also gotten customer feedback to identify other pain points and have created additional software to address these.
Spotlight and Part ID which we will talk about in more detail later on in the presentation.
Lets take a look at our flagship products and then we’ll talk about how these are implemented in different industries.
UMA Smart Desktop – w/ or w/o 6 axis robot, under 12 inches
UMA Smart Station – 2 sizes, 2x4 and 3x6, 12 < x < 36 inches
UMA Smart Cell –greater than 36 inches
Configurable. Always has robot and scanner, sometimes rotary, sometimes a rail that the robot is mounted on, and sometimes both.
We work with a number of partners which we are a Certified systems integrator (CSI) for
Total solution provider
Bridge the gap between robotics or automation companies and 3d scanning companies.
Focus on accelerating productivity though the automation of quality control and dimensional inspection by using the latest technologies in robotics and machine vision.
Something I want you to keep in mind as we go through the presentation, is what are the problems we are addressing?
Problems addressed by using automated 2D or 3D inspection.
Contact measurement only inspects what you tell it to touch, with 3d scanning you get the whole picture
People make mistakes, 3 different operators inspecting the same part with the same tool will give you 3 different results. Then they will make mistakes writing it down or entering it into the PC, if they do this at all.
Quality managers tell us how much more quickly they can get parts out and they can now hold suppliers accountable on incoming parts. Reduce scrap.
Improved analysis of surface data with color maps
Inspectors constantly tell us how much more they enjoy their job now that they don’t spend all day climbing around large parts trying to scan them manually. They can focus on tasks they are better at.
Data is king. More data -> Better process improvements.
Here’s an overview of some of the types of companies we work with.
Most of these companies have deployed automated systems from 3D Infotech.
Of course, there are many others not on this list.
You can see the variety of industries: automotive, aerospace, heavy machinery, consumer electronics, healthcare
Dimensional Inspection
Sample reports.
You’re looking at a CAD model with some datums and other GD&T callouts defined on it that we want to inspect. GD&T (geometric dimensioning and tolerancing)
Color maps – Ranbow and Pass/Fail (easy to see how the part is behaving or where it is out of tolerance)
Statistical process control (SPC) on both surface data and features (trend charts and statistics such as Cp or Cpk
See engineers in a circle around a part scratching their heads, guessing at what could be wrong.
This gives them the data they need to make intelligent decisions.
Complex aerospace castings. (48 inches)
On a CMM can easily take 4-8 to inspect a part like this, but w/ 3d scanning we can have a more comprehensive report in less than 20 minutes
Less bottlenecks.
More data -> process improvements.
E-stop, UR10, rail, rotary, scanner
Pass/fail maps. If the casting is bad, don’t waste any time machining
Material handling across sites, if it’s bad, don’t waste money shipping.
Sierra Alloys, ATI, Arconic
Oil industry
Laser scanning has come a long way in the past few years.
can scan black or shiny w/o spray or targets
Biggest drill bit is 8 meters (~24 ft)
Delivered cells with 8 or more meters in length. We’ve done cells this size for inspecting large aerospace panels and entire truck chassis.
Nobody else can do parts this length
What you’ll see in each of these pictures is an LMI gocator
Traditionally Gocator known for in-line high speed measurement. Here we have a slower offline solution that gives all the benefits of the traditional Gocator, such as accuracy, affordability, built in temp comp, no calibration req.
Gocator value add – Pif file creation, automatic scan to scan alignment, scan to CAD comparison, more advanced GD&T.
Royal technologies making plastic parts for the automotive industry. You can see our case study on our website.
Thermostat cover – excellent GR&R results.
Gage repeatability and reproducibility – GR&R testing and correlation studies offered.
Pain points we are addressing:
Time to set up a CMM program
Time to run CMM program
Fixturing issues
– can set up generic path for small volume measurement. One robot pose with a shot every 45 degrees. Non-contact measurement is much more forgiving in terms of fixturing.
There’s a couple ways we are helping out companies using 3d printing.
The first is inspection of parts coming right out of printer. Each part is different.
Layer by layer inspection (in-process inspection), built software to make this easy
Picture in the top right. Scan data is being cleaned for printing or to be turned into a CAD model.
Give our customers the ability to do reverse engineering themselves, but also offer as a service if it’s a more infrequent task.
Adaptive Manufacturing
Using sensor information to drive secondary processes.
Picture in the bottom right, there are some areas out of tolerance. We are finding where the out of tolerance areas are and creating a tool path for a robot to grind down the out of tolerance areas.
We’ve done similar projects for drilling applications where we are drilling new holes based on the locations of the current holes.
Large project combining CNC and CMM tending. Load part to CNC, when it’s done, place it on cmm, feedback loop to correct tool offsets based on cmm results, automated binning based on pass/fail.
Now Lets take a look at some ways we’ve extended our solutions that customers are taking advantage of.
CLICK ON THE VIDEO
The first is with augmented reality. We call our software Spotlight AR++.
Video on top shows manual workflow.
Bottom shows projection tracking which we’ve deployed with some of our Smart Cells.
After a part has been scanned, we are projecting dimensional information back onto the part.
in this example, we are using a sheet metal part.
Useful for manual reworks such as marking, grinding, welding, masking, or shimming operations. Guide the operator for where to rework.
We can also display features in a step by step presentation mode. Aerospace companies using this for assembly guidance. Virtual work instructions eliminates the need for physical templates and reduces error.
We can use any projector and support the use of multiple projectors. Preferred over tablet or glasses because it wont get lost/damaged/stolen, and doesn’t create a barrier between the human and the workpiece.
Guided part placement.
If you’re an operator placing parts to be scanned, rather than developing custom fixtures, you can use a software we’ve created to place your part accurately. We call this software Part ID. Now lets watch the video
This system is at Metalsa who is scanning sheet metal parts in free state. Because the robot is programed with the part in a certain location, if we place the part in the exact same spot, we can ensure all critical data is captured everytime and get better repeatability.
Now we’ve seen many applications of 3d scanning for dimensional inspection. Lets take a minute to see the two main software components that make this possible. The first of these is Streamline, which is our automation software and the core of the automated solution.
Solves the problem of having to integrate all of the hardware together.
Removes need for robot programming or any other custom programming.
Plug and play solution for connecting different pieces together to offer simplified HMI for the operator.
Helps deliver automation projects quickly while providing traceability and many other key functions.
All of the data analysis you have seen so for has been done in PolyWorks. We are the master distributor of PolyWorks on the west coast.
This is an extremely powerful tool that has become a standard in many industries, especially automotive and aerospace.
Built in SPC. Macro scripting capabilities. GD&T
Other useful tools not shown here: thickness maps, airfoil gages, flush and gap gages, cross sections. Depending on what industry you’re in.
PolyWorks for CMM. Much easier to create a measurement sequence than write a program in another software.
Universal software that works with any 3d metrology device.
Let’s see how all of these two pieces of software and the different hardware components come together in our Smart Station.
Let’s move onto defect detection. This is Non-destructive testing or NDT.
How are companies in metals industries are currently doing this? What happens is the part is coated with a dye penetrant which fluoresces under UV light. Then, a person in a dark room with special lighting will manually look for defects.
The camera you see here is ML2000 – UK (different from typical machine vision camera/3D scanner).
Can scan in any lighting conditions, so it can be used in-line. Smaller foot-print, removes operator.
Useful for finding cracks, pits, positives that typical metrology device might not see.
Software allows you to classify/quantify what is a defect. Rule based so less subjective.
Ability to assign clusters to a group of islands.
Here we are able to classifying welds as good or bad using deep learning.
What is deep learning
Example-based training modeled on human learning
Not a rigid rules-based solution
Neural network comes up with own analysis and compares to what the operator told it was good and bad
The Operator makes some adjustments as necessary, until the vision system’s understanding is the same as his
Inspection of weld seams to check for pin holes or overlap
Not possible without deep learning
Medical Applications using deep learning
Imagine training a standard vision system to find these defects on the threads and back of the medical screw while ignore the reflections
Medical screws need to be prefect. The challenge is that the parts is not flat and has a highly reflective texture.
Using rules based methods is impossible due to countless variations and countless potential defects
Training with only good images : 6 sets of 10 images of good screws
Here is an application for assembly verification of large aerospace assemblies.
In this example we are looking at brackets which have an inconsistent orientation because they are assembled manually.
We are checking for several different conditions seen here. Good, screw missing, bracket completely missing.
This can be useful on other components as well components such as lockwire, clamps, potting. These types of assemblies can have dozens and dozens of items we need to check for.
We were approached by several companies for this type of application because they are spending thousands of dollars to ship parts worth six figures, only to have them failed and sent back because their operator had overlooked something as small as a missing thread cap.
In this case, a customer is assembling computer mice and needs to verify that all the components are present.
Because of their inconsistent shapes, it’s a challenging application for traditional machine vision.
Even if traditional vision could solve it, you’d need different solutions for each component type – much harder to maintain in the field.
Why 3D Infotech?
Knowledge of automation and inspection – have worked with SME to very large companies.
UMA platform - Adapt to different part geometries, volumes, finishes, cycle time or accuracy requirements. Future proof.
Upgradeable and scalable with other hardware or software such as part id or spotlight
The value add we bring to other sensors such as Gocators to auto-generated .pif files, utilize better meshing algorithms, manage scanner settings such as resolution or shutter speed.
Perhaps most importantly, eliminate custom solution challenges – unknown development and integration costs, maintenance and support, rapid deployment, continuous improvement
Recap:
3 types of inspection – dimensional, cosmetic, and assembly verification
Bottom line, we can reduce scrap and alleviate bottlenecks
I hope you all learned something today.