Reducing risk when purchasing a vision system

Purchasing a vision system solution is often a CapEx cost to your organization. This article highlights a strategy for risk reduction of this purchase.  

Eliminating risk!

So we have to ask, are you an electrical engineer, a computer scientist, a physicist or a mechanical engineer?  

Remember the old joke, what is the difference between Mechanical Engineers and Civil Engineers? Mechanical Engineers build weapons, Civil Engineers build targets.

Why do we ask? Machine vision experts are usually one of the above but to be proficient at their trade, they have to be a jack of all trades. This is expertise that is normally acquired over a career. Learning how to deal with substrates, optics and vibrations in equipment. The issue that most manufacturers face is they task a project engineer with the task of purchasing a vision system and possibly other complex machines. That engineer will not have been exposed to enough vision projects during his career to gain that knowledge. To reduce risk it's key to build a team. This team should have industry experts that have that experience so as to support that project engineer.

The next step once the team is in place is to plan. So let's define the goals and steps that you need to take.

1. Build a Vision Map. Understand the best location to install the system and any other types of vision systems.
2. What are the key requirements as to performance?  Have you been able to get a complete sample set tested? Are you absolutely clear on the resolution required, the optical setup and the algorithm to be employed.
3. What is the correct technology to use? Is the vendor pushing some canned product that he can sell for the best gross margin? Turnkey solutions with a fixed price are often the preferred route, but remember that the vendor is then incentivized to use the cheapest hardware to optimize their profits. Understand what you are purchasing. The key is vendor independence through the use of commercial off the shelf technology. ( COTs)
4. Machine vision companies are 80% software and 20% system engineering. Think about open formats and if the platforms that they are proposing will support 3rd party software integration. That's the first question you would ask if purchasing a software platform. 

Common shortcomings when buying a vision system

Think about how machine vision systems have been used and purchased to date. The vision system has been bought for one purpose only, to inspect a product on a line and to alert the operator of any defects. This is usually in response to a customer complaint. This is very much analogous to security systems in the home. Often we do not take action until we have had an incident. So very little forward planning has been done.

The inspection system as used is dedicated to alert major defects to an operator. If an inspection system is on a line where the product can be rejected in real time then corrective action may not even be taken until almost all product is rejected. Very often the purchase of an inspection system is done as part of a larger project and is deemed a add on. There is often no thought how it is going to integrate with companies wider operations. The equipment acts as an island used by the operator for that moment on that line.

The return on investment is based on reduced customer complaints and if used at the point of production (this is not always the case) a reduction in inline waste.  If an inspection system is used to inspect material on a final process for defects introduced upstream it will not reduce waste. It may reduce customer returns. The net result is that the vision system is seen as a cost

Wrapping Up

The message is simple. It's all about planning. Fail to plan, plan to fail!  The key is to expand your team so that you have experts at hand, especially in the early stages. Vision systems are basically IT solutions. When building an IT solution, most companies have clearly defined plans and goals, and then work with a close partner that has the domain expertise to implement that plan. They rarely segment the system into parts and offer each element to the lowest bidder, who then supplies the cheapest hardware to do the job.