You’re a bunch of players

Having discussed the pros and cons of vision system standardization, (“Do you play the field?” November 14^th,) I conducted a poll to find out just how fickle you lot are.

The answer is, very.

Less than a third of you use only a single make or brand of vision system, and some 40% have three or more different brands. This is interesting, and perhaps a little disappointing.

To be blunt, it says to me that you buy what the salesman pushes. When the guy from Keyence shows you their latest-and-greatest, you buy it. And the same for Cognex. And Microscan. And Banner. And so on. I may be wrong on this, though I doubt it. And it’s a pity. You’re giving yourself substantial learning curves, spares, and support issues and making life more complicated than it needs to be.

Now the number of respondents was pretty small, so I may be being unfair on some of you, but really, three or more different brands? Next time you’re shopping don’t just believe the salesman, ask me instead.

Cross-shopping

In the automotive world Sales and Marketing types like to talk to about cross-shopping. The three series BMW is often cross-shopped with the Audi A4, the VW Beetle with the Mini. In other words, consumers are smart enough to check out competitive products before plunking down their hard-earned.

So here's my question: how often do you cross-shop vision systems?

I'll bet the answer is “practically never.” I'll bet that when you need a smart camera you call your friendly local electrical bits-and-bobs distributor and ask what he can offer. If he carries Cognex, that's what you'll get. If he works with Omron, or Panasonic, well guess what you'll be using.

Is this a smart way to buy machine vision? I don't think so.

I'm shopping for a new car right now. I figured out my needs, (interior space, good gas mileage, good warranty,) and my budget, and now I'm compiling a spreadsheet where I can compare the models that meet my constraints. The final decision will still involve some subjective criteria – how it looks, how it feels – but I'm comfortable with that because I know I'll be working from a base of quantifiable data and will be making an informed decision.

I think we should buy vision systems the same way. There are a lot of vendors out there with products that differ but all have strengths and weaknesses. So before you buy a Matrox Iris, a Banner PresencePlus P4, or a Cognex InSight, figure out your needs and see which fits best. That way you won't be buying a Maserati when what you need is a minivan.

Machine vision needs its own Steve Jobs

A common theme among the many eulogies for Steve Jobs was his gift for simplifying what already existed, making devices such as mp3 players and phones a delight to use.

Oh how we need some of that thinking in machine vision. I’ve just spent a couple of days learning to use some new software from one of the leading vendors, and boy was it painful!

The actual vision tools are great; it’s just that the whole interface is overly complicated and clunky. Even the trainer was having difficulty finding the functions he wanted. I have no doubt that once the learning curve is climbed I’ll find it all very powerful. But I don’t recall my iPod having a learning curve. That, in my humble opinion, was the genius of Jobs: he made stuff so elegantly simple that it became a joy to use.

The only machine vision software products that have come close to being as intuitive as iTunes are the old DVT Framework product (remember that?) and the Banner vision software. Yes, they both had severe limitations but they made it easy for a guy in a factory to quickly set up an application.

You may argue that I’m nit-picking, but I disagree. If I’m going to use a product five days a week, fifty weeks a year I can put up with a learning curve. But the engineer who’ll implement one or two systems a year needs it to be intuitive and simple.

Vision software companies: heed my call. If you want to sell more software, make it easier to use.

Vision sensor upgraded

I have always been a fan of the vision products from Banner, so I paid close attention to news of an upgrade to the iVu sensor line. This product has been around a year, maybe longer, but it doesn’t seem to have attracted much interest, so let me give you a summary.

The iVu is a vision sensor, much like the Checker from Cognex. It’s a small box with integrated lens and designed to run just a single vision tool – perhaps an intensity check or a pattern-match. What I felt gave the iVu an edge however was the small display screen on the rear. This means a user can see what it’s doing without needing to hook up a laptop.

A weakness with the iVu was the absence of any Ethernet connectivity. This made it a little awkward to get it talking to other production equipment. Banner must have agreed with me because they’ve now launched the iVu Plus. This has, and I’m quoting from the Banner website, “Ethernet communication, multiple inspections and a new Sort Sensor.”

I haven’t got my hands on the iVu Plus so I can’t say just how easy and capable it is, but if it follows in the footsteps of its upscale brethren, the PresencePLUS line, it should be a user-friendly product. I also have no word on pricing but my guess is that it’s in the $1,400 - $1,700 range, so it’s definitely a Checker alternative.

“You cannot be serious!”

Well that seems to have been the response to my Banner iVu - Cognex Checker 3G comparo. At least I know people read what I have to say, even if they don’t agree with me. So let me defend my position.

I accept the point that the rear-mounted screen on the iVu may be difficult to use in every installation, and I can hook up a laptop to the 3G, so that’s a point to Cognex. But while the laptop will help with initial setup, it’s not a permanent visual indication that the system is operating. If you’ve read my earlier posts about the Banner cameras you’ll know I’m a big believer in the value of seeing a live image while the system is running. That makes the Cognex quite a bit more expensive.

I also accept that the 3G can run at a really, really high frame rate, but I’m not convinced that that this compensates for a lack of pixels. In fact I don’t think anything compensates for a lack of pixels. When it comes to imaging, pixels are your raw data. You can’t add data that isn’t there; interpolation is just taking a guess at what might be present.

As I said previously, I do believe, (although this is based more on faith than any real facts,) that the Cognex algorithms are superior to those from Banner. If the 3G had the same resolution as the iVue, Cognex would be the clear winner, but it doesn’t and they’re not.

Come on Cognex. If you want to be the Champ you can’t just go around saying you’re the Champ; you’ve got to deliver a knockout blow. So give us a VGA 3G and I’ll call you the winner.

Banner iVu versus Checker 3G

In the vision sensors arena these are the two newest contenders. Both claim to be easy to deploy, so let’s take a moment to compare them.

The iVu, from Banner, is priced at $1,200 while the Cognex Checker 3G is $1,000. But the iVu has a higher resolution sensor and provides a live display of what the sensor’s ‘seeing’ via an LCD panel on the rear. In that regard it’s a lot like a consumer digital camera. You’ll need to drop an additional $600 with Cognex to get a live view from the 3G via their “SensorView” interface pendant, but you can hook it up to multiple 3G’s.

One little quirk to watch is that there are different model references for the iVu, depending on whether the outputs are PNP or NPN. A small but significant detail.

Without doing a direct, hands-on test it’s difficult to compare the imaging tools, but I suspect the 3G has the edge. Cognex have so much strength in software development that’s it’s hard to imagine they’d put an inferior product out in the marketplace.

So which would I buy? Of course, it’s task-dependent, but on balance I think I’d go with the iVu. In my opinion the superior resolution compensates for the less finely-honed algorithms. Plus, I really like having a live display on the back on the sensor. That’s been a gripe of mine with the Insight range, and while SensorView is an option it’s more money and needs hooking up.

The win goes to Banner, but it’s very close.

Excellent guide to machine vision lighting

From ‘Vision & Sensors’ mag comes a first class primer on illumination for machine vision. In this lengthy and detailed article Brent Evanger of Banner Engineering discusses the importance of lighting, as well as the options available, (LED, fluorescent, halogen,) and lighting geometry, (darkfield, brightfield, backlighting.) It’s a well written piece with good information for both the vision ‘newbie’ and the experienced pro.

And while I’m praising the article, let me sling more praise in Banner’s direction. It’s no secret that I’m a fan of their low end, inexpensive, vision sensors. Yes, they are limited in their functionality, but the Banner software is easy to use – in fact almost intuitive – and the price is right. If you’re considering a vision system from Keyence or Omron you should also look at Banner. Check them out at http://www.bannerengineering.com/en-US/products/5/Vision/14/PresencePLUS-Vision-Sensors.

Let’s talk about it

When it comes to debugging non-functioning equipment, (hardware and software,) I have a theory. This states that I will never be the first person on the planet to have encountered this problem, and therefore the answer is out there somewhere, usually accessible on the World Wide Web.

A number of machine vision vendors have also recognized this, and have set up discussion forums to facilitate the sharing of information. (I suspect it also reduces the load on their Tech Support staff.) The best known of these is probably the DVT SmartList, (run by Cognex since their acquisition of DVT,) although it does have a weakness: it’s not searchable, which must mean that the same questions keep popping up over and over.

National Instruments also have an excellent discussion forum, which is searchable. And if you want to take the road less traveled, for NI disciples there’s also the LAVA forum.

On the other hand, Banner, whose vision products I like very much, offer no such online support. That’s a pity, and I suspect it may also be costing them business. One of the factors in the success of NI, (and I know there are many,) is the way in which their users have become an online community, sharing experience and helping the less knowledgeable.

Banner guys – please set up a discussion forum. It will help me, and it might help you too.