Machines are getting smarter by the day. However, we’re not talking about tragicomedies like Honda’s stair-fumbling robot or the uncanny valley “Sophia” calls home. The machines we’re interested in are the more than 75 billion connected devices and assets that will be online by 2025. This Internet of Things will change the world as we know it once 5G, machine learning and machine vision all get folded in.
The 5G network will provide unprecedented data transmission speeds throughout the industrialized world. Machine learning, and eventually AI, gives us the means to analyze huge quantities of data and optimize human enterprises far better than we could unassisted.
What about machine vision? In machine shops and other manufacturing environments, it’s just as important as these other Industry 4.0 technologies. If machine learning is next-level cognition for sensors, robots, tools, and other hardware, then machine vision is next-level environmental awareness.
Machine vision requires one or more high-definition cameras to function. These cameras scan the operational area of the machine, or a workpiece already placed in it, and make judgments about its type, size, position, orientation and condition.
On an automotive parts assembly line, for instance, using machine vision might involve installing stationary cameras to automatically illuminate and capture an image of each part as it moves through the assembly stages. On-board analytics then receive these images and quickly compare them to known parameters, effectively automating the quality assurance process.
The important feature of this technology isn’t the camera, but rather the underlying logic and algorithms interpreting the images as they are captured. This is where machine learning and machine vision converge. The more images that are processed, and the higher the quality of those images, the more accurate the results become over time as the machine better “understands” what it’s looking for.
Several types of lighting are common in machine vision systems, the choice of which depends on the nature of the process and the qualities of the working environment. The right amount and intensity of lighting must be accounted for. Then, machine vision uses a combination of pixel counting, edge detection, pattern recognition, character recognition, color analysis and metrology to make accurate and higher-order judgments about an object or location’s physical properties.
Here are some other ways machine vision is already in use.
Machine shops, no matter their size, are regularly called on to deliver quick turnarounds, tight tolerances and better-than-industry-standard quality. If you’re reading between the lines, you can already imagine some of machine vision’s implications for delivery estimates, quality assurance and cost-effectiveness.
Here are some of the current applications of machine vision in machine shops, as well as small- and large-batch manufacturing facilities:
Many companies report positive business outcomes after automating quality assurance processes. Inspection stations powered by machine vision add another layer of confidence compared with human inspectors, who are susceptible to burnout after just a few hours of this kind of repetitive work.
Even better, lots of companies have found ways to automate in this area without losing employees.
Future workplaces, especially in heavy industries, will be populated by flesh-and-blood employees alongside new types of robots and cobots. Machine vision is critical to ensuring these mechanical assets navigate their work environments safely and carry out their jobs without disrupting nearby employees.
Pathfinding using machine vision can be as simple as painting a line on the floor for an automated vehicle to use or as sophisticated as an autonomous order-picker confidently navigating warehouse racking.
Speaking of warehouse racking, machine shops and many other industrial environments rely on ongoing counting, sortation and transportation of raw materials, partial workpieces and finished products. Machine vision is an important addition to material handling processes.
As mentioned, it can allow a cobot to pick items from warehouses for order fulfillment with superior accuracy. This is valuable in any industry that relies on timeliness, accuracy and low error rates.
Opportunities abound in inventory management and counting tasks, too. Machine vision systems can tell at a glance whether a correct number of items is accounted for and whether they are in the expected condition.
A major advantage of machine vision is that assets equipped with this technology can apply logic and analysis to the images they capture right at the source. This is compared with sending pictures to an external server for review, which is potentially expensive and a drain on digital resources.
Machine learning and vision push our analytical potential further out to the edge of our industrial networks. The cloud and the Internet of Things delivered distributed computing. With higher-order logic available, a great deal of that data can stay at rest rather than being transported or stored.
Maintaining product throughput and synergy between departments is a challenge in machine shops and other manufacturing settings. As cameras and pattern recognition become more common in industrial IoT systems, they’ll find further usefulness in helping coordinate facility activities and assets. The larger the field of view, the greater the scalable potential.
Drones and other unmanned implements in industrial settings can perform ongoing analyses on the health of infrastructure and traffic throughout the work environment. They can also help engineers and managers better target their efforts in addressing maintenance tickets and process bottlenecks. Our economies require greater efficiency at every level, and machine vision lends us the eyes and ears we need to achieve it.
Other uses for machine vision across manufacturing and other industries include testing and calibration, data collection, real-time machine monitoring, process control, security and much more. As board-level image recognition and higher-order logic becomes affordable and accessible enough to deploy at scale, there’s plenty of room for innovation when it comes to self-aware and self-correcting machines.
As for machining shops, they’re no stranger to answering difficult job specifications and design briefs. Ensuring consistent quality in a smaller or medium-sized machine shop might be doable with traditional, hands-on methods. These days, though, if any operation hopes to scale larger, some element of automation is probably required.
Any machine shop that wants to cut costs and improve the speed with which they tackle their work should look into machine vision. Automation is best applied to repetitive, high-risk activities first — like quality assurance and the others covered here.
Article by —
Megan Ray Nichols
Freelance Science Writer
Read More article
Your email address will not be published. Required fields are marked *
Copyright © 2015. Divya Media Publications Pvt. Ltd. All rights reserved