Edge AI: Revolutionizing Real-Time Data Processing and Automation

From smart home assistants like Alexa, Google, and Siri to advanced driver assistance systems (ADAS) that can alert drivers of lane departures, the world relies on edge AI to provide real-time processing capabilities for these increasingly important devices. Edge AI uses artificial intelligence directly on the device, performing computations close to the source of the data, rather than relying on cloud computing in remote data centers. Edge AI brings lower latency and faster processing speeds, reducing dependence on a constant internet connection while reducing privacy concerns. This technology represents a major shift in the way data is processed, and as the demand for real-time intelligence grows, edge AI is well positioned to continue to have a strong impact in many industries.

The biggest value of edge AI is the speed it brings to critical applications. Unlike cloud/data center AI, edge AI doesn’t send data over a network link and expect a reasonable response time. Instead, edge AI computes locally (usually on a real-time operating system) and excels at providing timely responses. For situations like applying machine vision on a factory line and determining within a second whether a product can be diverted, edge AI is perfectly capable. Likewise, you don’t want the signals from your car to rely on the response time of a network or cloud server.

Edge AI for real-time processing

Many real-time activities are driving the need for edge AI. Applications such as smart home assistants, ADAS, patient monitoring, and predictive maintenance are all noteworthy applications of this technology. From quick responses to home issues, to vehicle lane departure alerts, to blood sugar readings sent to smartphones, edge AI provides fast responses while minimizing privacy concerns.

We’ve been seeing edge AI working well in supply chains, particularly in warehousing and factories, for quite some time. The technology has also grown significantly in the transportation industry over the past decade, with delivery drones that can navigate through conditions such as clouds. Edge AI also offers huge benefits to engineers, especially in medical technology, an area that is critical for advancement. For example, engineers developing pacemakers and other cardiac devices can provide doctors with tools to look for abnormal heart rhythms while also proactively programming the device to provide guidance on when to seek further medical intervention. The medical technology sector will continue to increase its use of edge AI and further advance its capabilities.

Generate Edge AI Model

As more and more systems in our daily lives now have some level of machine learning (ML) interaction, understanding this world is critical for engineers and developers to plan for future user interactions.

The biggest opportunity for edge AI is ML based on pattern matching based on statistical algorithms. These patterns can be the presence of a person being sensed, someone just saying the "wake word" to a smart home assistant (like Alexa or "Hey Siri"), or a motor starting to shake. For a smart home assistant, the wake word is a model that runs at the edge without sending the voice to the cloud. It wakes up the device and lets it know it's time to give further instructions.

There are several ways to build machine learning (ML) models: either using an integrated development environment (such as TensorFlow or PyTorch) or using a SaaS platform (such as Edge Impulse). Most of the "work" in building a good ML model is creating a representative dataset and labeling it well.

Currently, the most popular ML models in edge AI are supervised models. This type of model is trained on labeled and tagged sample data, and the output is a known value that can be checked for correctness, just like a tutor grading homework. This type of training is often used in applications such as classification work or data regression. Supervised training is very useful and has high accuracy, but it relies heavily on labeled datasets and may not be able to handle new inputs.

Hardware for running edge AI workloads

At DigiKey, we are well positioned to help enable edge AI as it typically runs on microcontrollers, FPGAs, and single-board computers (SBCs). DigiKey works with top suppliers to provide multiple generations of hardware that can run ML models at the edge. This year we’ve seen some great new hardware launches, including NXP’s MCX-N series, and we’ll soon be stocking ST Microelectronics’ STM32MP25 series.

In the past few years, development boards from the maker community have been popular for running edge AI, including SparkFun's edge development board Apollo3 Blue, AdaFruit's EdgeBadge, Arduino's Nano 33 BLE Sense Rev 2, and Raspberry Pi models 4 or 5.

Neural processing units (NPUs) are increasingly being used in edge AI. A neural processing unit is a special-purpose integrated circuit designed to accelerate the processing speed of ML and AI applications based on neural networks. Neural networks are based on the structure of the human brain, with many interconnected layers and nodes called neurons that process and pass information. A new generation of NPUs is being developed with specialized math processing capabilities, including NXP's MCX N series and ADI's MAX78000.

We also see AI accelerators for edge devices, a space that is still being defined, with early companies to watch including Google Coral and Hailo.

The Importance of ML Sensors

High-speed cameras with ML models have been playing a role in the supply chain for quite some time. They have been used to decide things like where to send products in a warehouse or to spot defective products on a production line. We are seeing vendors developing low-cost AI vision modules that can run ML models to identify objects or people.

Although embedded systems are required to run ML models, more products are being introduced as AI-enabled electronic components. This includes AI-enabled sensors, also known as ML sensors. While adding an ML model to most sensors will not make them more efficient, there are several categories of sensors that can be greatly improved through ML training:

• Camera sensors to develop ML models to track objects and people in the frame

• IMU, accelerometer, and motion sensor for detecting activity

Some AI sensors come pre-installed with ML models that are ready to run. For example, the SparkFun evaluation board for people perception comes pre-programmed with face detection and returns information via the QWiiC I2C interface. Some AI sensors, such as Arduino's Nicla Vision or Seeed Technology's OpenMV Cam H7, are more open and require a trained ML model for what they are looking for (defects, objects, etc.).

By using neural networks to provide computational algorithms, objects and people can be detected and tracked as they enter the camera sensor's field of view.

The Future of Edge AI

As many industries grow and rely on data processing technology, edge AI will continue to gain wider adoption. By enabling faster and more secure data processing at the device level, innovations in edge AI will be far-reaching. Several areas we see expanding in the near future include:

1. Dedicated processor logic for neural network algorithm calculations.

2. The advancement of low-power alternatives compared to the huge energy consumption of cloud computing.

3. More integrated/module options, such as AI vision components, will include built-in sensors and embedded hardware.

As ML training methods, hardware, and software continue to advance, edge AI is poised to grow exponentially and support a wide range of industries. At DigiKey, we are committed to being at the forefront of the edge AI trend, and we look forward to supporting innovative engineers, designers, builders, and procurement professionals around the world with a wide range of solutions, seamless interactions, powerful tools, and rich educational resources to make their work more efficient. For more edge AI information, products, and resources, please visit DigiKey.com/edge-ai.

Shawn Luke is a Technical Marketing Engineer at DigiKey, recognized worldwide as a leader and continuous innovator in the commercial distribution of electronic components and automation products, offering more than 15.6 million components from more than 3,000 quality name-brand manufacturers.

By Shawn Luke, Technical Marketing Engineer at DigiKey