Perform Machine Vision at the Edge for Factory Automation
September 08, 2022
Machine vision runs the gamut in factory automation applications, from handling security and maximizing production, to providing predictive maintenance for the machinery. On the production side, tasks could include inspection, orientation, identification, and assembly. While each of these elements could be handled by a human, there are many (emphasis on “many”) reasons why you would want to do this with a machine/computer.
Let’s start with accuracy. It’s obvious that the machine vision system is far more accurate than the human eye. It can operate in speeds that a human cannot. Think about products continually coming down the production line, 24/7. The machine vision system doesn’t take breaks, except for maintenance, doesn’t get tired or sick, and doesn’t require days off.
The steps taken in the machine-vision architecture include:
- Capturing an image
- Process the captured images and apply corrections when needed
- Analyze the data from the processed images
- Make decisions/take actions based on the data analysis
- Alert key personnel as necessary
- Retain data for future comparisons and on-going analysis
Machine vision can also be used to implement what many in the industry are calling Industry 5.0, which refers to the concept of employing collaborative robots, also known as cobots, in a manufacturing environment. The collaborative nature of the cobots means that they can operate safely and securely on the same floor as humans. To accomplish this task, machine vision is required.
Choose the Right Camera for the Application
One thing that’s almost certain in a machine vision application is that a lot of cameras are required. The resolution and placement of those cameras is dependent on the application, and in most cases, they will be varied throughout the floor. For example, the camera that handles the product inspection will have to be far higher resolution than the one that is used for security purposes. While the former needs to pick out defects in a product, the latter just needs to know when a person has entered an environment.
This all ties back to the Edge computer. It’s obvious that all interactions must be synchronized back to the Edge of the industrial IoT (IIoT) for the simple reason that there is no time to go back to the Cloud to make decisions. When products are rolling off an assembly line, the potential latencies incurred going to a Cloud computer would not halt the production line fast enough when needed, potentially causing irrevocable damage. In the case of the cobots, the same principle applies—a decision is needed immediately to circumvent harm to a human.
Powering Your Machine-Vision-Enhanced Facility
To power a machine-vision architecture, the Edge system must have the appropriate compute capability. An example of such a computer is the WINSYSTEMS’ ITX-P-C444 Arm-based SBC, which is well suited to handle AI/machine-learning applications. It consumes less power compared to competitive, meaning that it’s suited for remote field locations. The ITX-P-C444 fits the industrial Pico-ITX form factor. It’s designed with NXP’s i.MX8M applications processor and offers dual Ethernet, industrial I/O, and other expansion options.
(The WINSYSTEMS ITX-P-444 SBC takes advantage of a low-power CPU, based on the Arm architecture. It’s designed to the industrial Pico-ITX form factor.)
A second offering from WINSYSTEMS to consider is the PX1-C441 single-board computer (SBC). It’s designed to a PC/104 form factor, but don’t let its small size fool you. It’s powered by the latest generation Intel Apollo Lake-I dual- or quad-core SoC microprocessors. In addition to up to 8 Gbytes of soldered down LPDDR4 system memory and a non-removable eMMC device, the SBC features PCIe/104 OneBank expansion. Further, the PX1-C441 supports M.2 and SATA devices.
(The WINSYSTEMS PX1-C441 SBC can handle the machine-vision capabilities of your manufacturing facility.)
In addition to the Edge-based machine-vision application, the PX1-C441 is also suited for industrial IoT applications and embedded systems in the industrial control, transportation, Mil/COTS, and energy markets. That is thanks to its small size (4.55 by 4.39 in.), rugged design, and extended operating temperature range (-40°C to +85°C).
Edge Versus Cloud
Keeping your operations at the Edge of the IIoT is great for control purposes. This means that the Edge computer can decide what decisions need to made immediately versus those that can be parsed out to the Cloud for further analysis. Predictive maintenance is good example of what goes out to the Cloud. That data is analyzed over time, meaning days, weeks, or even months, to determine when a machine should be brought offline for maintenance. In most cases, there’s enough lead time to take the machine offline at a time when production will not be impacted.
The experts at WINSYSTEMS can help you determine what should be handled at he Edge versus the Cloud, what type and how many cameras should be deployed, and so on. Realize your future through machine vision.