The potential automated inspection project has been identified. A thorough evaluation of needs, budget and return on investment has been performed. All specific application requirements have been clearly defined and communicated. A functional specification has been developed which defines project scope, inspection tasks, automation requirements and acceptance criteria. This machine vision project is at the critical point where someone actually must make it work.
Systems integration for a machine vision project is the process during which significant value is added to components in order to create a complete inspection system or automation cell ready for use in an industrial environment. To be specific, one definition of integration is “the combining of parts so that they work together to form a whole.” In the case of a machine vision application, those parts, or components, will include lighting, optics, cameras, processors (computer or smart camera) and perhaps a variety of other automation technologies such as PLC controls or robotics. Successful integration for an automation project involving machine vision requires not only significant technology-specific skills, but also competency in a variety of engineering areas. Let’s outline the tasks facing the systems integrator.
System SpecificationThe integration process starts with system specification. Up to this point, all analysis, documentation and specification have been focused on the “what” and “why” of the desired inspection. The system specification details the “how.” Armed with an application analysis and functional specification, the integration team must first evaluate imaging and lighting for the project. For any machine vision application, this is the most critical step.
The correct definition of lighting and imaging technique, and the specification of the correct illumination, imaging and optical components needed to acquire a usable image has a greater impact on the success of the vision application than any other integration task. System specification continues with identification of automation issues, including part presentation and image acquisition. Often, one of the most difficult parts of an automated inspection application is to ensure that the part or parts being inspected are presented to the cameras in the correct orientation-with little or no positional variation-and that the control system can initiate the inspection at the correct point in the process. Particularly in gaging applications, and to a lesser extent in applications involving assembly verification, variations in the position of features within the image can significantly affect the reliability of the inspection.
DesignA thorough system specification should provide detail on types of components, particularly in the case of imaging and illumination, and should state all positions of the cameras and lights relative to the part. From this information, a complete design of the machine vision station or inspection machine will be developed. It is from this design that the system will be fabricated and built. The complexity of the design is dependent on the scope of the inspection project. In some cases, only simple mounting structures might be provided for cameras and lighting on existing automation equipment. It also is possible that a complete and complex machine may need to be built.
With respect to the layout for imaging, it is always important to design adequate adjustment to the position of the cameras and lights. Too much adjustability, however, can make the system difficult to set up and maintain. The design will include a full bill of materials that specify all components to be used for the project.
BuildUpon approval, the system design can be fabricated and built. Ongoing review and quality oversight during this phase can help to catch design or implementation issues that might cause problems or waste time later.
Software Configuration, Programming and TestingWith few exceptions, no matter what the application, a certain amount of configuration or even programming must be done to make the machine vision component perform a reliable inspection. Unless the application is extremely trivial, the integrator must have a very good understanding of the target algorithms that will be used to extract and analyze image data, and must be adept at implementing and configuring those algorithms on the machine vision component platform chosen for the project.
It would not be an understatement to say that many machine vision applications, even those that have been properly specified up front, suffer in reliability or fail due to incorrect implementation of the tools available within the machine vision processor. Furthermore, it is at this point in the project where deficiencies in component specification might be realized. Many inspection applications ultimately are, or become, more complex than originally thought. Sometimes components with too little capability or resolution are forced into applications that require significantly more power, algorithmic depth, processing speed or pixels.
A good rule during system configuration or programming is to perform constant testing of the theoretical inspection concept or the entire inspection system where appropriate. If possible, it helps to have a good collection of stored images captured from the actual production environment. By using this type of sample set and testing process, more of the normal production variation can be accommodated in the system prior to installation and validation.
System Installation and ValidationFinally, the system is ready to install. Scheduling becomes a key issue, particularly in situations where production must not be significantly impacted. Again, having a high confidence level in system operation prior to installation is important; system development on-line in the production environment is never efficient and promotes a lack of confidence in the final implementation. The process of validation varies widely, but should be clearly stated in the original functional specification. The inspection criteria and deliverables should be tested as specified and documented. There should be accommodations made for regular in-process validation of the system.
Taking on the Machine Vision Integration ProjectAfter examining the previously mentioned steps, it is time to take on the project. There are five key ingredients needed at the launch of a machine vision project: a champion, management support, a functional specification, acceptance criteria and an organized team.
The ultimate success of a machine vision application also requires:
• A realistic budget
• Systematic evaluation of vendor capabilities
• In-house machine support
• Input from production staff
• Clear communications with the vendor
• Support of vendor with latest drawings, parts and specifications
The champion is a person who is committed to putting the personal energy into the project to see that it is successful. This person has good leadership skills, and is allowed the time required to fully implement the project. Experience has shown that without a champion, any project is in severe risk of failure. The champion must get management support for the project.
The key ingredients for this support on a machine vision project are that it must make good investment sense, it must push a corporate hot button, and it must have a good chance for success. Management support is indicated not only by having a financial commitment, but also through active participation by management in monitoring the project and helping free up needed resources.
When applying machine vision, senior management should:
• Foster an atmosphere that encourages change
• Support change agents
• Demonstrate buy-in to change
• Encourage plant and line to take ownership
• Establish and support realistic schedules
Recognizing that no single person possesses all the skills and knowledge needed to successfully evaluate or implement a machine vision application, a project needs a team with a complete set of skills. While the entire team does not need to be assembled for the launch, part of the project plan is to identify where these skills will come from when needed. Important skills are project management, process engineering, electrical or mechanical engineering, optics, software, electronics, operations, training and maintenance.
Most of these skills are needed intermittently during the project. Some of these people will come from the organization that will be the ultimate user of the machine vision system. Other skills may come from a systems integrator or consultant.
Considering an Outside Systems IntegratorThe question of whether an end user implementing an inspection system needs to partner with a machine vision systems integrator is a function of 1) whether the company feels comfortable enough with the technology to undertake an installation using internal resources, and 2) whether those resources have the time to do justice to such a project, or whether that time could best be spent doing something else that is more valuable to the company.
The latter question is particularly important. Machine vision technology is not beyond the reach of a competent engineering staff. However, the time involved in the integration process for machine vision may not be a good economic investment for a company. Systems integrators will significantly accelerate the rewards achieved using machine vision. Their application skills, having been engaged in many applications previously, allow them to quickly identify the correct complement of camera, lighting and staging. With their understanding of commercially available machine vision products and the application, integrators typically can guarantee using a system that provides the best price vs. performance trade-off.
An end user should ask several questions to determine whether or not a systems integrator should be used:
• Do I have ready access to all the skills this project requires?
• Can I maintain the skills I develop and benefit from the experience of integrating a vision system?
• Do I have sufficient time available to see the project through to success?
• If the project turns out to have unexpected challenges, how will I see they are met?
• Am I able and willing to retain ownership of the system for the long term, to ensure maintenance, support and upgrades?
If an end user can answer all these questions satisfactorily, then they may be able to handle the system integration task. However, in most cases, the answer to one or more of the questions is not satisfactory. In those cases, an outside integrator should be used.
Working with a systems integrator entails an implementation alliance. The in-house staff provides process knowledge, usability and commissioning support. The integrator provides cost-effective labor, application expertise and ongoing support. The component manufacturers provide product knowledge, component expertise and lifecycle support. Each party can focus its abilities on issues that may be difficult for its counterparts to handle effectively. If managed properly, with ongoing communications maintained between all the parties, this implementation alliance can significantly increase chances for success.
Systems integrators provide efficiencies in system design, fabrication and installation, and provide the required training. With an in-house project team in place, taking ownership of the system after installation generally goes smoothly. Having a systems integrator involved in a project provides an excellent resource for ongoing application support and for future system improvements if required. For success in a combined implementation approach, the scope of work must be clearly defined for each party, and the communication channels must be clear so that minor adjustments made by one party do not cause unexpected work for another. In a successful alliance, accountability must be defined. V&S
David Dechow is president and founder of Aptúra Machine Vision Solutions (Lansing, MI). For more information, call (517) 272-7820 x11, e-mail firstname.lastname@example.org or visit www.aptura.com.
Tech TipsHere are some questions an end user should ask to determine whether or not a systems integrator should be used:
- Do I have ready access to all the skills this project requires?
- Can I maintain the skills I develop and benefit from the experience of integrating a vision system?
- Do I have sufficient time available to see the project through to success?
- If the project turns out to have unexpected challenges, how will I see they are met?
- Am I able and willing to retain ownership of the system for the long term, to ensure maintenance, support and upgrades?
How to Choose a Systems Integrator PartnerAs the entry barriers for doing business as a machine vision systems integrator continue to diminish, more companies say that they integrate machine vision. While this competition is good for the marketplace, the consumer must be wary of those companies that are not dedicated to long-term, broad support of this challenging technology, and that just do some machine vision integration as a side business. Here are some criteria for choosing a systems integrator in order of importance:
- The integrator demonstrates the ability to be a good partner.
- The integrator has available the necessary set of skills needed by the project and the required technology.
- The integrator has demonstrated that he can deliver a working system in a reasonable time frame.
- The integrator has shown he will be around for the long term to support the equipment and the relationship.
- The integrator has experience with similar applications.
- The integrator has experience in this or similar industries.
Finally, the integrator should be willing to offer a guarantee of system performance. The guarantee will be relative to the prepared functional specification and should be contingent on the stated acceptance criteria. Such a guarantee establishes the integrator’s confidence in the application, and his ability to successfully complete the integration.