UKIVA Chairman, Allan Anderson, says introducing a vision system into your organisation may seem daunting, but breaking it down into simple steps can make the task much easier.
Introducing industrial vision into your manufacturing operation seems daunting. As with anything, industrial vision has a learning curve – which can be taken step by step. With the correct approach, even those new to vision can sidestep basic mistakes – and start the process with their eyes open.
It’s helpful to look at a vision system as a puzzle with seven pieces. All seven need to fit together when developing a working vision system. These pieces are algorithms, complexity, flexibility, set-up, acceptance cost and knowledge.
System software – An algorithm is the brain of any vision system, the software program that underpins it. Three of the most commonly used types are pattern matching, analysis, and character recognition and verification. Between them, they cover many vision system applications.
Pattern matching locates a part or object in a field of view – and verifies its presence (or absence). It can also help guide robots and detect whether all pieces of a larger part are present. A typical application might be on a pick-and-place line.
An analysis algorithm is for applications such as counting pills in a foil packet. It can identify properties such as contour, elongation and centre point. It can be adapted to handle both simple and complex applications.
Character recognition/verification identifies printed characters – which is not always easy for a machine. A variation in background, lighting and other factors can affect performance. In some cases, an advanced algorithm is needed.
Cutting complexity – Vision systems are typically used by non-specialists, not vision experts. So, they must be easy to understand – and simple to use. A key factor here is a clear graphical user interface (GUI). Outputs must be clear and recognisable. For example, whether an item has passed or failed should be obvious. This is typically communicated by either a green (pass) or red (fail) box. Even in complex, fast-moving environments a GUI should present critical information (such as past results, component images and camera selector) in a clear, simple format.
Flexible approach – Vision systems will need ongoing changes and maintenance. Sometimes, this requires experts being called in to make small, simple adjustments. This can be avoided by building flexibility into the system – such as by using flowchart-based machine vision software. If you’re skilled enough, you could build the vision system yourself. Alternatively, employ a vision expert to do it.
Once the system has been deployed, learn how to make changes and edit the flowchart. There is no coding involved, just blocks of a flowchart – which should make things less intimidating.
Set-up – Set-up is critical – and cannot be rushed. Remember that vision systems often have contact with production areas. So, all hardware, sensors, cameras, illumination and cables must be correctly placed and properly connected. Cameras will then trigger properly from their correct positions – giving the vision system the best chance of performing well.
It’s also important to verify the type of I/O that’s being used, as the two main ones – Polling I/O and Real-time I/O – work in quite different ways.
Getting buy-in – Sometimes in manufacturing, a vision system has been championed by the engineering team – but not by production. Resistance can then grow from a sense that the system is a competitor for those jobs. This can lead to it being used incorrectly, switched off or even damaged. This is why buy-in is critical: a vision system’s value must be accepted across the company.
Cost equation – In manufacturing, cost is critical. A vision system can set you back less than £10,000 – or more than £100,000. Analyse factors such as return on investment in advance. Determine how much it costs to make your products – and how much failures cost you. This allows open discussions to take place with a vision system supplier – so that expectations are clear and cost targets are met.
Knowledge boost – Finally, note that you can learn to create your own vision system. There are many courses available – though it does take time and dedication. If this isn’t something you can manage, UKIVA’s many machine vision experts are on hand to help.
Looking carefully at the seven pieces of the puzzle should help ensure that you have the best possible vision system for your needs.
