In machine vision, put the emphasis on usability
A discussion with Mike Lawn, ICS Inex Inspection Systems
A discussion with Mike Lawn, ICS Inex Inspection Systems
Mike Lawn is director of applications engineering at ICS Inex Inspection Systems, Clearwater, FL, USA; www.inexvision.com. He has been involved in the vision industry for more than 15 years, working with established systems for the glass-container and beverage industry, as well as leading new business development in other markets. Editor in chief Conard Holton spoke to him about trends in the design of inspection systems.
VSD: What is your primary line of business, and what important developments do you see in machine-vision technology and products that drive this business?
LAWN: I have been fortunate to have spent most of my working career in a company that is one of the oldest and most experienced machine-vision businesses. Our original systems, designed for the beverage industry in the 1950s, used up to four photocells and a spinning prism to scan bottles for contaminants. They performed and, in some cases, still perform admirably.
The introduction of CCDs and PCs opened a new era in machine vision. Since then, processors have become faster and faster, cameras have higher and higher resolutions, and algorithms are more and more sophisticated. Systems developed today are full of advanced functionality and megapixels, but at a cost. Sales have become feature driven, regardless if they are needed or a real benefit. I see end users struggling with complex setup, a multitude of parameters rarely used, and confusing terminology.
As with many consumer products today, we seem to have forgotten that technology is intended to help us. I believe the trick is reining in the technology and concentrating on usability: harness the latest technology to create intelligent algorithms that need a minimum of setup with maximum practical performance, concentrate on a simple coherent user interface, and hide the technology behind understandable controls.
Our old bottle scanners used four sensitivity dials to control the inspection performance. Our aim is to continually advance the performance and capabilities but get back to that level of simplicity-and maybe only one dial.
VSD: What market changes are driving these developments?
LAWN: Our core business is glass-container inspection. We provide a range of equipment that detects and rejects faulty ware. The majority of the faults occur in the forming stage, yet our systems are not inspecting until the final stages of the process. When something goes wrong, there can be extremely high levels of waste, with up to an hour of already formed production affected.
In recent years our developments have concentrated on systems that detect defects earlier in the process or prevent them from being created at all. Our Mini-Lab product is designed to automatically sample sets of containers from the production line and perform a series of highly accurate tests. These measurements are monitored with SPC software for control variations, where actions can be taken to prevent the problem drifting to an out-of-specification state. In this way, quality can be maintained or improved, efficiencies increased, and costs minimized.
We are learning that this form of process control is not only extremely attractive to manufacturers, but is becoming a minimum requirement for future equipment.
VSD: You have offices in North America and Europe; what differences do you note between machine-vision technologies and business developments in the two regions?
LAWN: As a recent immigrant from the UK, I am often asked this question. Ten years ago I would have been able to reel out a long list. Today there is so little difference. The business world has been global for sometime, and the rest of life is rapidly catching up. The machine vision business is no different. I see the same technologies, the same business plans, the same customer requirements and needs. I do notice slightly more exposure to inspection technology in Europe than here in the USA. The majority of our new customers in the UK have the wariness and realistic expectations associated with previous experience, while here in North America it is more often the opposite.
VSD: How has the advent of low-cost smart cameras impacted your development and deployment of inspection systems?
LAWN: As a company we have always dipped our toes into other markets to develop and diversify the business. The advent of smart cameras appeared to allow us to accelerate this process. With the flexibility of these systems we could develop new systems in a fraction of the time and at minimum cost.
In reality, as an equipment manufacturer providing turnkey solutions, aggressive DIY [do-it-yourself] marketing prevented us from being able to compete and still provide the level of service we demanded. Smart cameras have brought vision to the masses, and brought immense opportunities for skilled independent integrators as end users realize that producing a long-term reliable solution may not be as simple as it first seemed.
Smart cameras may be the future for all vision systems, but I personally have my doubts. Mini PCs and digital cameras provide more flexibility with software options and potentially better long-term support for hardware.
VSD: How will the impact of Microsoft’s .NET framework and the soon-to-be-released Vista operating system affect system designs of the future?
LAWN: Not long ago we were struggling with 200-MHz 286 PCs, DSP pipes, and assembly code. Now we run 3-GHz Dual Xeon Duo-Core processors utilizing technology from the gaming world (see discussion of the Gemini 3-D system in Vision Systems Design, Dec. 2005, p. 18). There is no doubt that without Microsoft’s ‘Bloatware’ and Sega none of this would be possible. Still, as a platform for machine vision, we use Linux whenever possible.
Mini-Lab sample-measurement system developed by ICS Inex Inspection Systems is designed to prevent defects rather than merely rejecting them. It uses high-resolution Camera Link cameras and Zeiss optics.
VSD: How do you use Ethernet within the automation of your systems? What developments do you see occurring in deterministic Ethernet protocols such as EtherCAT, Ethernet-IP with determinism, and the isochronous real-time ProfiNET-IRT?
LAWN: Our current vision systems use 1000M Ethernet to communicate between the various PC-based modules. A combination of UDP and TCP messages are used to transfer images, configuration files, and tracking messages.
In recent years I rarely leave a customer meeting without talking about remote communication links. Data collection, centralized control, and SPC are “the buzz” in the manufacturing industry as they strive for improved quality and efficiencies. Machine-vision systems are ideally suited as a tool, and we should all be taking note.
We use the OPC protocols as an alternative to propriety formats to transfer this type of data to external monitoring and control systems over Ethernet.
Automation protocols have always presented a problem for those of us who export around the world. Europe, North America, and Japan each have their own preferences on control platforms and associated protocols. The OPC Foundation (Scottsdale, AZ, USA; www.opcfoundation.org) has developed an open standard that is becoming generally accepted and supported by most of the major players. Hopefully, it will supersede propriety protocols for basic control communications.
Our real-time I/O control is currently confined to direct I/O. With high-speed applications, accurate triggering and reliable encoder-based tracking is essential. Until recently, remote I/O has not been practical or cost-effective on our standard equipment. The new deterministic and other real-time Ethernet automation protocols look like they will become a viable solution in the next few years.
VSD: Object-oriented programming tools are now allowing more sophisticated machine-vision software to be deployed. What do you see as the main advances in this area?
LAWN: As a project manager and not a professional programmer, I see the advantages of building a robust object-orientated framework. It invokes thought into the ultimate aims of a project and controls the structure of future code. This type of structured programming improves the ability to efficiently portion out coding and helps simplify long term support.
VSD: What developments are you seeing in lighting and illumination, and which choice is best for each application?
LAWN: For high-speed applications, xeon strobes still can’t be beaten--so much white light in such a short time. For slower-speed applications, LEDs are the way to go for now: nonhazardous voltages, longer life, and more controllable. Current developments are pushing the luminance issues, but they are not here yet. My tip for the future would be the nanotube-a maturing technology with excellent potential.