Vision pioneers

by Andy Wilson, editor
[email protected]

It’s always interesting to visit trade shows to see the latest products, technologies, and applications, especially if you get paid to do so. While attending these shows, I always like to interrogate the staff of component manufacturers and system integrators regarding their products and technologies to understand more than may have been documented on a company’s web site or datasheet.

During the interview, it is often necessary to use off-the-cuff journalistic tricks to prod the person being interviewed into revealing more than they would normally. After patiently listing to company representatives, I like to throw in my favorite catch phrases such as, “That’ll never work” or “I wouldn’t have done it like that” to measure the person’s reaction. Being involved in a high-technology industry, of course, it is likely that once in a while I will be correct in my assumptions.

But, as many a purchaser of machine-vision products will know, often a component being advertised does not quite match up to the performance promised by the salesman on the booth. Or, although a machine-vision system is shown to be working on the show floor, when it is installed in a factory, other factors (most often lighting) render the machine inoperative.

Some years ago, after I had uttered one such catch phrase, I was asked my opinion. After I explained the philosophy behind my reasoning, the company representative explained that, while my ideas may have been more elegant, cost played an important role in the design of the machine. Because of this, the company in question had developed its own image-processing software, coding simple algorithms such as edge detectors in C++ that have for a decade been part of numerous off-the-shelf packages. And, rather than use standard off-the-shelf frame grabbers to capture images, the company had also developed its own camera interface boards and coded the algorithms in on-board FPGAs.

Of course, all of this cost time and money but, I was told, having such a proprietary solution allowed the company to offer a cost-effective system that was unrivaled in performing its particular task. Initially, the end user of the product was happy since the product performed as engineered.

However, due to the recent recession, the system integrator responsible for engineering the equipment was forced to lay off personnel and, sadly, had to declare bankruptcy. This occurrence was especially upsetting for the end user since when parts finally failed and the algorithms needed to be updated, no one could provide support.

On the frontier

Although the system integrator may have pioneered new low-cost technology in the development of the system, the lack of any off-the-shelf parts or software meant that once the system failed it could no longer be upgraded.

Customers looking for machine-vision inspection systems do not have to face such problems, even if their system suppliers are relatively small companies. For example, if a specific lighting product is required to perform a particular application, customers could do worse than contracting with a major lighting supplier to develop the component—rather than have the system integrator build the product. Similarly, the use of off-the-shelf machine-vision libraries would ensure that even if a particular integrator were forced into bankruptcy, software support would still be available.

Indeed, cost may be an important consideration in the development of a machine-vision system. But end users should realize that, although greed may initially appear to be good, a thorough analysis of the components used by the system integrator may be more beneficial in the long run. So, the next time you visit a trade show seeking advice from a system integrator, be sure to use my catch phrases to dig a little deeper than just what the salesman quotes you for a price.