Artificial Intelligence of Things (AIoT) adoption is growing – further proof that two proven technologies can have greater value when used together. 

Manufacturers and their supply chain partners are sold on the Internet of Things (IoT). The technology is becoming ubiquitous in the industry, driven in part by advancements in wireless technology that enable fast and reliable transmission of the data these systems generate. Although not as widely adopted as IoT, artificial intelligence (AI) is also increasingly meeting manufacturers’ and distribution centers’ needs to extract value from large and complex data sets.  

The two technologies are an obvious complement to each other. An Industrial Internet of Things (IIoT) system’s network of sensors, devices, computers, and systems generate and exchange massive volumes of data. AI enables computers to learn, solve problems and make decisions, similar to the way humans do. AIoT systems, which combine the two technologies, promise advantages, including enabling automation, predicting equipment malfunction, and providing decision makers with insights they can leverage to optimize processes.  

The potential of AIoT is driving market growth, expected to rise to more than $65 billion by 2025. Investments in AIoT are seeking returns that neither technology can do on its own, such as: 

• Facilitating the use of unstructured data, including video or audio, in analysis 

• Reducing latency by powering automated systems that generate and use data at the edge  
• Immediately and accurately predicting next best steps in a process by analyzing real-time and historical data 

AIoT in Action 

Use cases for AIoT are virtually limitless – it can be applied anywhere where an operation can leverage data analysis to inform, automate, or optimize. A recent application involved using AIoT to power healthcare robots during the coronavirus pandemic. These robots performed tasks, such as disinfecting facilities, sifting through recyclables, and transporting materials in hospitals, so that people didn’t have to put themselves at risk of contracting COVID-19. In addition, AIoT helped monitor crowd levels in high-traffic areas to inform residents of places where social distancing could be difficult.  

AIoT also helped manufacturers meet the challenge of protecting their employees during the pandemic. With AIoT remotely monitoring machinery, analyzing data including vibration, speed, temperature, and pressure, and taking necessary action at the edge, staff could safely keep their distance and monitor activity remotely.  

Another successful AIoT application is our smart pallet system. With warehouses under pressure to find ways to operate more efficiently, quickly and accurately, this AIoT system automates pallet stacking and asset location. It can intercept packages placed on pallets in error before they are shipped incorrectly, keep inventory up to date in real time, reduce waste and unnecessary shipping costs, and create a safer work environment.  

Perhaps one of the most publicized AIoT use cases is the role it plays in autonomous vehicles. The AI model that powers the vehicle’s control system will work in concert with IoT infrastructure that provides cars with data about the weather, traffic, routes, obstructions and other vital information.  

Making AIoT Work 

Effective AIoT systems require a robust network to support them. As communications systems evolve from 4G to 5G, edge computing will provide the speed, reliability, low latency, and increased capacity to support the hundreds or thousands of IoT devices generating data. 

Edge appliances also vital AIoT system components as they process the most critical data at the source for lower latency and greater reliability. Edge computing can make a system more economical, especially if the costs of transmitting data to the cloud from a remote location would be cost-prohibitive. Edge computing can also be the solution for operations that, for security or regulatory compliance reasons, cannot send all data to the cloud. We like to categorize edge computing that supports AIoT into three primary buckets: AI on Modules (AIOMs), application-ready edge platforms (AREP) and service-ready edge platforms (SREPs). Here is a brief overview of what they are and how they work.

To build the AIoT system itself, enterprises can use AI on Modules (AIOM), the AIoT parallel to the computer on modules that combine CPU, GPU, and VPU for heterogeneous computing. AIOMs from ADLINK include embedded products, such as COM Express, SMARC, and CompactPCI.  

Rugged application-ready edge platforms (AREP) are exactly what they sound like- edge devices that have been designed and built for specific industry applications to help end users and system integrators with edge computing. For instance, the AVA-5500 is an application-ready edge platform that is ideal for use in transportation. When it comes to railway digitalization for instance, the AVA-5500 can be used to support railway hazard and intrusion detection, automated train operation, and train delay prediction which are all AIoT applications. AREPs are designed to be used in any environment, from a bumpy train, to areas where the platform can be exposed to cold, wind and rain, to remote or underground locations. AREPs from ADLINK include AI Smart Cameras for AI-based machine vision, edge inference platforms, and real-time ROS 2 robotics controllers, as well as high-end GPU servers and multi-access edge computing platforms. AREPs deliver high performance AI inferencing, high integration capability with AI software and systems, making it an easily serviced and cost-effective solution.

Another option is service-ready edge platforms (SREPs), cost-effective, low-risk, vendor-agnostic solutions with modular architecture. They provide real-time data connectivity and processing at the edge from any data source to any data source, allowing machines and controllers in industrial environments to take real-time action. ADLINK SREPs include the ADLINK Edge™ software platform, ADLINK Edge™ apps with out of the box connectivity for developing various proprietary IoT services and applications, and ADLINK Edge Smart Solutions, which are industry-specific hardware and software integrated solutions including Smart Pallet and Machine Health solutions. 

Putting It All Together 

With the right hardware, software, network, and system design, enterprises can create systems that provide the best of both the IoT and AI worlds. Once all the components are in place, enterprises can leverage AIoT to break new ground, find new growth opportunities, create a stronger competitive position through automation, and reach goals they’re setting for a more efficient and profitable future – today.