Skip to main content

Computer Vision at the Edge Speeds IoT Development

edge, AI, computer vision

As more industries realize the value of connected machines and devices, IoT adoption is growing by leaps and bounds. By 2025, there will be more than 25 billion connected things in industries ranging from electricity and gas to retailing, wholesaling, and transportation (Figure 1).

Bar chart with projected IoT device growth numbers
Figure 1. The number of IoT-connected devices will exceed 25 billion by 2025. (Source: Statista)

The IoT gives businesses real-time information they can use to solve problems quickly and operate more efficiently. But traditional IoT technology has some significant limitations. Because it uses wide-area networking to transmit data to the cloud for processing before returning results to the decision-maker, it does not deliver the real-time decision support needed for warehouse robotics. It also requires huge bandwidth resources.

These shortcomings have given rise to new approaches that combine computer vision technologies with artificial intelligence.

Computer vision—aka machine vision—enables machines to identify objects and analyze scenes and activities in real-life environments. When paired with artificial intelligence, computer vision can simulate the human capability of solving complex problems. This Industry 4.0 approach drives industries from automation (where humans program machine decisions) to autonomy (where the machine makes decisions based on real-time data).

A More Powerful Edge Solution

One company that excels in providing its customers this advanced capability is ADLINK, whose success comes from merging artificial intelligence, machine vision, and high-speed data connectivity at the edge.

Founded by Jim Liu in 1995, the Taiwan-based global company uses embedded computing technology to deliver the Analog-Digital LINK (hence the name) required to accelerate deployment of artificial intelligence at the network edge. This approach improves operations and services across multiple industries.

Daniel Collins, ADLINK Senior Director of Edge Solutions, says, “Edge computing is about applying the right data, at the right time, in the right place, to drive the right decision and take the right action. Accomplishing that outcome demands the employment of artificial intelligence at the edge.”

The Smart Pallet solution couples #MachineVision and #ArtificialIntelligence to provide fully automated visibility of all #warehouse packages and pallets. @Adlink_IoT via @insightdottech

Improving Warehouse Operations with Computer Vision and AI

An excellent example of effectively deploying computer vision at the edge is the ADLINK Edge Smart Pallet solution. This product is an ecosystem of warehouse technology, including applications, sensors, analytics engines, and operational systems.

Connected through the ADLINK Data River, these components combine to deliver autonomous Industry 4.0 solutions to address multiple customer pain points at the edge, such as:

  • Packages that contain the wrong merchandise or are lost, stolen, or delivered to the wrong place.
  • Lack of inventory visibility during packaging, palletization, and distribution.
  • Siloed and unconnected systems that make automation and process monitoring impossibly complex, intrusive, and expensive.
  • Bandwidth and latency challenges posed by transmitting data to and from the cloud.
  • Privacy concerns associated with some 5G WAN solutions.

The Smart Pallet solution couples machine vision and artificial intelligence to provide fully automated visibility of all warehouse packages and pallets. This methodology replaces traditional barcode hand-scanning with machine learning, helping to automate pallet stacking and box finding across warehouse facilities. It also eliminates inventory errors caused by misplaced and non-scanned items (Video 1).

Video 1. The ADLINK Smart Pallet solution uses machine vision and AI to determine whether warehouse packages have been loaded onto the right pallet. (Source: ADLINK)

By capturing multiple image data streams and applying high-performance processing power, machine learning at the edge can increase warehouse automation while simultaneously improving efficiency and accuracy.

Edge computing systems also eliminate the need to transmit data to the cloud, using a high-bandwidth local area network to collect and process data locally. Smart Pallet connects to new and existing equipment, using a vendor-agnostic approach built on open standards. The entire platform is architected with modular components to ease integration with existing IT and OT systems.

Creating the Machine Vision Solution

One of the biggest technical challenges ADLINK had in developing the solution was collecting enough data to train the artificial intelligence models. It solved that problem by using Intel® technology in the computer vision cameras.

The solution’s standard package consists of an ADLINK NEON-1000-MDX, industrial-grade smart camera with onboard storage and compute. That component uses an Intel Atom® processor, an Intel® Movidius Myriad video accelerator, and the Intel® OpenVINO Toolkit machine vision framework—which comes with pre-trained AI models. Using this framework as a starting point, ADLINK created its own AI models to locate individual packages and provide a visual display of the pallet on a computer monitor.

As a longtime Intel partner, ADLINK also relies on Intel’s supply chain security. This aspect is of critical importance to customers who are required to meet legal and regulatory certifications.

Boosting Efficiency and Stopping Theft

Automatically reading barcodes saves enormous amounts of time for warehousing and manufacturing operations. After adopting ADLINK, a global meat processing factory reduced scanning time by 90% and improved overall processing speed by 41%. The plant also significantly reduced labor costs and scaled its business to create $340,000 of additional revenue.

Another company discovered the value of the smart pallet solution after shipping customers 500 empty packages—which were supposed to contain cell phones. It turned out they had all been stolen.

To prevent this type of loss from ever happening again, the company installed the ADLINK system and began using computer vision to monitor boxes and pallets. The deployment enabled them to detect and prevent theft during the packaging process, leading to a dramatic reduction in product shrinkage.

By blending machine vision and artificial intelligence at the network’s edge, ADLINK has established itself as an industry leader in bringing analog data into the digital world. Once companies have its solution in place, they can easily customize it for new uses.

“Everything is modular,” Collins says. “You just plug in the building blocks you need and extend them to something else. The possibilities are endless.”

About the Author

Kevin L. Jackson is a globally recognized Thought Leader, Industry Influencer and Founder/Author of the award winning “Cloud Musings” blog. He has also been recognized as a “Top 5G Influencer” (Onalytica 2019), a “Top 1000 Tech Blogger” (Rise Social Media 2019) and provides integrated social media services to AT&T, Broadcom, Ericsson and other leading companies. Mr. Jackson’s commercial experience includes Vice President J.P. Morgan Chase, Worldwide Sales Executive for IBM and Engility Corporation Director Cloud Solutions. He has served on teams that have supported digital transformation projects for the North Atlantic Treaty Organization (NATO) and the US Intelligence Community.

Profile Photo of Kevin Jackson