The Mercedes Benz chapter in Ralph Stein’s 1967 book The Great Cars, opens with a description of the importance of Daimler and Benz in building the world’s first horseless carriages. The chapter continues with the story of how Wilhelm Maybach designed the first Mercedes ─ the car that made, “every other car in the world obsolete.” Here Harel Boren, CEO of Inspekto, founder of the Autonomous Machine Vision category, explains how a similar change is about to shake up quality assurance.
Visual inspection was originally performed by a human operator by hand and eye, much as carriages were drawn manually by horses. However, the introduction of machine vision digitalized the process, acting as the horseless carriage of inspection. Unfortunately, the machine vision industry is problematic and flawed ─ it is awaiting the introduction of the Mercedes, in the form of Autonomous Machine Vision, to revolutionize quality assurance (QA) and make all current offerings obsolete.
To reduce the risk of defective products, manufacturing best practice currently involves continuous investment in machine vision QA equipment. However, there’s a long way to go before any QA is done. QA managers must assess each point on the manufacturing line, to determine whether the efforts and investment associated with placing a Machine Vision QA solution at each particular position is viable. This is because of the long wait, high costs, and extensive downtime associated with the design, preparation, installation, and commissioning of these solutions.
Challenges and problems
Traditional Machine Vision is limited by the shortcomings of image registration, computer-vision, deep learning, and other artificial intelligence (AI) technologies. The inherent limitations of these technologies have predetermined the capabilities of today’s Machine Vision solutions. As a result of these inherent limitations, planning and installing existing technology is a complex and lengthy project that requires the expertise of a systems integrator.
At present, a manufacturer must hire an integrator to identify possible issues, create Proof of Concepts (POCs), and select the exact components that will make up the solution, from an endless variety of lighting, cameras, lenses, and other components. Additionally, the manufacturer must invest in integration to bring this together as a hard-engineered solution on the production line.
Because this process requires such a high level of expertise, its development is completely inaccessible by a plant’s personnel ─ the manufacturer is entirely dependent on a systems integrator. In turn, the QA Machine Vision ecosystem has naturally aligned itself to serve the systems integrator ─ the only party equipped to deliver a functioning Machine Vision QA solution.
On top of all this, the manufacturer must shut down the production line while the integrator installs and commissions the equipment, causing further downtime, loss of income, and increasing the cost of the project.
The challenges don’t end there. Any change on the production line, however minor, requires the assistance of the integrator to adjust the solution to function correctly. If the manufacturer modifies the line, perhaps to produce a different item or to monitor another stage in production, the Machine Vision solution will not adapt itself. Instead, the manufacturer must once again call in an integrator ─ usually after defects have gone undetected for a while ─ to adjust the solution or, worse, to produce a new one from scratch. Whatever change is made, the manufacturer is at the mercy of the integrator’s agenda, prices, and schedule.
Finally, the roller coaster of costs and efforts imposed by setting up traditional tailored Machine Vision solutions limits where they can be placed and what they can be used for. As a result, QA managers are compelled to place Machine Vision inspection solutions only at main junctions in the production line.
Finally, any points on the production line before a main junction continue producing defects that remain undetected for several steps on the line. If the manufacturer was made aware of the defect as soon as it was introduced, it could scrap the defected component before wasting more energy or effort on using it in as part of the manufactured product. With Traditional Machine Vision, this is impossible. The cost, time, and effort to build a solution acts as a blockade to increasing the number of inspection locations. This dire state of affairs will not, and cannot, change, as long as the technologies driving Machine Vision require a systems integrator.
The question is why must QA managers suffer through this for each visual QA spot? Why does the manufacturer deserve such a complex and expensive fate? Surely the industry is ready for a step forward.
Introducing Autonomous Machine Vision
The dawning of the age of Autonomous Machine Vision renders all previous Machine Vision technologies obsolete, just as the Mercedes once did all other vehicles. The technology grants industrial QA managers the independence and the ability to install inspection systems themselves, liberating them from the shackles of the systems integrator.
Autonomous Machine Vision puts the industrial plant and the QA Manager where they deserve to be; at the forefront of the industry. The systems comprise modern detection, acquisition, inspection, and self-integration capabilities, enabling QA managers to install, set up and operate an inspection system in minutes.
Not only are manufacturers given their independence from the integrator, the reduced cost, simple installation, and learning capabilities of the system gives the QA manager the freedom to install a system at any stage of production, or even move it from one place to another as needed. Gone are the days of operating Machine Vision solutions only at main junctions. We are now in the era of Total QA, where every stage of production can be valued equally, allowing inspection to identify defects as soon as they occur, to improve efficiency and prevent resource waste.
Quality assurance is a vital manufacturing process and traditional, integrator-centric Machine Vision solutions have run their course. Just as Stein’s book, The Great Cars, explained how the Mercedes would make all other vehicles obsolete, Autonomous Machine Vision is about to run the dependency on the systems integrator into the history books, giving rise to self-propelled, self-installed, self-tuned, and integrator-less Machine Vision QA. Industrial QA Managers are finally at the helm of QA performance on their production lines. Industrial manufacturing will never be the same again.
Filed Under: AI • machine learning, Industrial automation