AI • IOT • NETWORK EDGE

Machine vision applications promise greater efficiency, safety, and profitability—particularly for the industrial and smart city sectors—due to their ability to enhance inspection and quality control.

In factories, for example, automated optical inspection (AOI) can reduce manufacturing errors and increase productivity. And vision-based systems in smart cities can provide safer streets and better urban traffic control. But despite the broad range of potential use cases, these solutions can be difficult to implement.

“Industrial and outdoor urban environments are harsh, making it difficult to deploy industrial machine vision solutions in those settings,” says Kevin Lee, Senior Business Development Manager at Portwell, an industrial computing specialist that manufactures compact IPCs for machine vision applications. “In addition, there are tremendous demands for reliability and some strict space constraints in many industrial and smart city use cases.”

The good news is that modern embedded industrial PCs (IPCs), like Portwell’s WEBS-89I0, offer a computing platform that makes it possible to deploy machine vision solutions in even the most challenging scenarios. Rugged, flexible, and adaptable, these powerful edge computing platforms enable a range of new applications and already deliver value in multiple markets.

Embedded IPCs Unlock Machine Vision Benefits Worldwide

Portwell’s deployments in the EU and APAC regions are a good example of this.

In Japan, a large construction firm was looking for an automated solution to inspect and monitor building projects remotely. The company wanted to achieve technical oversight of field operations without the cost and inconvenience of sending an engineer or technician to the build for manual supervision. But the environmental conditions were challenging, with temperatures on-site ranging from 5C to 45C.

Modern embedded industrial PCs offer a computing platform that makes it possible to deploy #MachineVision solutions in even the most challenging scenarios. @Portwell_US via @insightdottech

Portwell helped the company set up a remote monitoring solution based on WEBS-89I0, its fanless box PC, which could withstand the rigorous operating environment while ensuring the reliability of the system. Cameras installed on-site would help to supervise operations to ensure that proper procedures were being followed and that the project was proceeding as planned, with the IPC doing the preprocessing and then transmitting relevant data to the company’s Microsoft Azure cloud for further analysis.

After implementation, the firm had achieved the level of oversight required, and no longer had to spend time and money sending skilled supervisors to the job site.

In a second Portwell deployment in the Netherlands, a system integrator was attempting to implement a smart city solution for a municipal government. The SI and local officials were concerned about safety and security on the city’s streets, and wanted to develop an automated surveillance system to detect dangerous situations and alert the authorities when necessary.

But due to the setting, the environmental constraints were extremely challenging. Reliability was also a concern, as the potential for equipment damage to an outdoor solution was high, and it would be inconvenient and costly for the SI to send an engineer to repair a computer in the field.

Portwell helped the system integrator develop a machine vision security system using its fanless embedded industrial PC as the edge computing platform. WEBS-89I0’s fanless design was chosen to reduce the probability of malfunction, since PC fans are the component that breaks most frequently when a computer is in constant operation. With this, a network of security cameras was set up around the city. Cameras were connected to the embedded IPC for edge analysis, with algorithms programmed to detect behavior that would raise an alert. The IPC’s built-in SIM card slot also made it possible to route data to a remote control center over local cellular networks.

Once deployed, city officials had their desired computer vision-based security solution—one that would require minimal maintenance and upkeep in the future.

Industrial Machine Vision: Flexibility and Reliability Speed Time to Market

Obviously, major differences exist between roadside traffic control systems, industrial AOI, and smart city safety solutions. The key to an embedded IPC platform that facilitates rapid development of diverse applications is flexible design and reliable, high-performance edge computing.

Portwell’s WEBS-89I0 embedded industrial computer, for example, offers a number of design features that make it easier for engineers and SIs to build for custom use cases.

Multiple USB and Gigabit Ethernet ports enable engineers to connect the WEBS-89I0 IPC to standard hardware devices like cameras; RS-232 and RS-485 ports offer extra connectivity for industrial equipment; and dual output ports provide a way to link the computer to displays. In addition, the computer’s compact footprint—a palm-sized 138mm x 102mm x 48mm—means it can be embedded into almost any solution without significantly increasing the overall size.

On the reliability front, Portwell’s technology partnership with Intel has been of great help in developing its embedded industrial PC. “Intel processors provide the balance of performance, stability, and energy efficiency needed to develop embedded applications,” says Lee. “Our partnership with Intel also gives us early access to next-generation processors, which helps us deliver market-leading solutions to our customers.”

For enterprises and SIs attempting to develop industrial machine vision solutions, this blend of powerful, reliable computing and flexible, adaptable design makes it easier to get to market faster—even when building highly customized solutions for buyers.

Collaboration Enables Wide Range of Industrial Machine Vision Apps

It seems likely that organizations in nearly every sector will look to incorporate computer vision technology into their operations in the years ahead.

In part, this is because implementation is now easier than ever, as modern AI technologies solve machine vision engineering problems more efficiently than older approaches.

“In the past, something like a factory AOI system for defect detection would have been tremendously complex to build using traditional programming methodologies,” says Lee. “But given the state of AI computer vision today, such a system can be designed and implemented far more quickly.”

In the smart city and industrial sectors in particular, availability of rugged, powerful embedded IPCs should help overcome adoption hurdles.

“Everyone in the smart city and industrial space wants machine vision applications because the business case is so clear,” says Lee. “But until recently, the biggest challenge was finding a suitable edge computing platform to implement those solutions. We believe we’ve overcome that obstacle.”

This article was edited by Christina Cardoza, Editorial Director for insight.tech.