• Font Size:
  • A
  • A
  • A

Feature Articles

Tackling the Challenging Vision Application

by Winn Hardin, Contributing Editor - AIA


Ask engineers about the commonalities of “challenging” vision applications, and they may launch into a discussion about a recent system with multiple cameras. Or they may relate another case where they fielded a vision system outdoors where lighting conditions are almost impossible to control.

Although both systems are very different, one being a complex system, the other a difficult challenge, both camps are right. Both complexity and difficulty contribute to the definition of a “challenging” vision application. But what does “challenging” mean, and how is it overcome? We asked that very question to a number of our industry’s finest integrators, and found a host of good answers and suggestions, from ways to manage expectations and manage mission “creep”, to the “anything, anywhere” system that has but one solution: walk away…

The Boundary Between Difficult, Complex
“Someone recently came to me with a machine vision system designed to find clear adhesive,” relates Perry West, president of the integration company, Automated Vision Systems Inc. (San Jose, California). “The problem had been addressed using polarization, but the material properties weren’t consistent, so the system was failing. Another application used multiple cameras with 21-axis motion to control an ink jet deposition process. The first was difficult because of the unique application, but the second was complex because of the high number of parts – both hardware and software -- that needed to be integrated. The applications were quite different, but both applications required a new approach to machine vision to solve the problem. Both were challenging.”

In other words, difficulty can come from pushing one aspect of a vision system to the limit – whether it is speed; contrast; spatial or temporal resolution; lack of constraints on product, imaging, environment, etc. Unconstrained customers can also lead to complexity. “Success can become its own problem as customers get excited and say, ‘the system is doing this, so it shouldn’t be too difficult to have it do that, too,” explains West. “To manage expectation creep, we try to formulate a 25-word sentence at the top of the specification that summarizes success. That way, when the customer tries to extend the system beyond the original design, we can point back to the specification that they approved and keep the project on focus. The summary statement can also be a useful tool for the engineer to sell the system to upper management. They don’t care about detailed specifications, only what the system will do for the company.”

One way to create a detailed ‘success’ statement is to focus on the parts, processes, and inspections that are close to the boundary of pass and fail. “Machine vision will invariably show the customer something they didn’t know,” explains Perry Cornelius, PE Advanced Systems Consultant, ABCO Automation Inc. (Greensboro, North Carolina). “It will show them a type of defect that may only occur once in 2 million parts. Sometimes the customers will want the vision system to find those defects, but we have to show them that finding that 1 type of defect in 2 million parts may not be worth doubling or tripling the cost of the system. When customers insist that machine vision find ‘anything, anywhere,’ the best solution is usually to walk away because you’ll never be able to make that customer happy. The customer and integrator should always be in agreement with the performance guarantees that the integrator has developed.”

Glenn Archer, Director of Business Development for EPIC Vision Solutions (St. Louis, Missouri), agrees that managing expectations is key to success, particularly when it comes to complex vision systems. “Everyone in this business has produced systems that worked, but the customer still wasn’t happy,” notes Archer. “A big part of avoiding that trap is managing the customer’s expectations of what can be accomplished with machine vision.”

Many Moving Parts
‘Difficult’ applications may or may not be complex applications that may or may not be solved by in-house engineers using commercial off-the-shelf (COTS) equipment. Complexity, however, born from integrating many systems and subsystems, is a sure-fire indicator of a challenging vision application.

“If you look at the breadth of machine vision projects out there today, I would guess that 60% to 70% could be solved and installed by a customer if they really wanted to use commercial machine vision products,” adds Archer. “With the other 30%, customers realize they need help. People who bring projects to us have usually either failed in their own efforts, or been turned down by other integrators who wouldn’t take the job, often because of the number of material handling and other automation technologies that need to be integrated with the vision system to make the solution work. We have all those disciplines in house, and that’s why we can succeed in these challenging vision applications.”

One of the most common characteristics of a challenging vision application is the need to interface the vision system to any number of pieces of automation equipment. “Challenging and complex vision applications often have to interface with material handling systems, PLCs, proprietary control systems, etc., – often in new ways,” explains ABCO’s Cornelius.

EPIC Vision recently built a system to inspect a bottle 360 degrees around its circumference without slowing down throughput. EPIC engineers adapted a filler machine by replacing six fillers with cameras, allowing each camera to capture an image showing 60-degrees of the bottle.  “Many vision integrators wouldn’t try to solve the application because of the unique material handling needs, and without a lot of industrial automation experience, you’re not going to be able to solve the problem,” says Archer.

Complex systems also require tighter integration of the engineers working behind the scenes. “One thing we do to mitigate the risk is to assign a project manager that oversees the mechanical engineer, electrical engineer, software developers, machine shop, and any other disciplines we require,” says ABCO’s Cornelius. “Some customers balk at having to pay for a project manager, but with systems of this complexity and cost, not having one person to track everything can lead to items falling through the cracks, or integration problems because one person doesn’t fully understand what another department is doing to solve the application. After building vision systems for years, the one thing I want to be able to do is sleep at night. To me, a successful vision system is one that works pretty much like a PLC. It does what we promise, without having to babysit it or have maintenance people tweak software or lighting every day - a system that can operate regardless of changes in finish, or what time of day it is. If we don’t think we can build a system like that, we don’t bid on the project, and customers appreciate that.”




There are currently no comments for this article.

Leave a Comment:

All fields are required, but only your name and comment will be visible (email addresses are kept confidential). Comments are moderated and will not appear immediately. Please no link dropping, no keywords or domains as names; do not spam, and please do not advertise.

First Name: *
Last Name: *
Your Email: *
Your Comment:
Please check the box below and respond as instructed.

Search AIA:

Browse by Products:

Browse by Company Type: