The global edge AI market is forecast to reach $107.47 billion by 2029, growing at a compound annual growth rate (CAGR) of 31.7% from 2022, according to Fortune Business Insights. Mainly driving this strong growth is the increasing application of AI in sectors including automotive, healthcare, personal computing, and industrial automation.

Indeed, AI has become indispensable in building a future connected world, where there are billions of things that are more secure, more connected, and more intelligent.

“We call this the Cloud Connected Intelligent Edge,” says Matteo Maravita, Senior Manager, AI Competence Center and Smartphone Competence Center, Asia Pacific Region, STMicroelectronics. “We believe that these devices will be much more autonomous and more connected to the cloud, thereby increasing not only the creation but also the local processing of data. All these devices cover several areas that we find in our daily lives.”

They can be found at homes, in factories, in workplaces, in cities and buildings, and in mobility solutions. “When we look at this wide variety of devices, and if we think about the related business model, we can understand how edge AI is going to be impacting our future a lot,” says Maravita.

Despite the continuous advancements in AI and machine learning (ML) technologies to enhance applications in many sectors, designers continue to face challenges—mostly hardware—when it comes to edge AI development.

“Basically, they want to implement AI, but they still need to look at the overall performance of the application,” says Maravita, He notes that designers need to ensure that the hardware they are going to use has a specific feature for security, for example, and to keep power consumption low.

Meanwhile, developers are also facing software challenges. For instance, they need to implement a learning model inside a machine. While they need to create, train, and deploy the model inside the final products, they also need to maintain this model. And this is a huge challenge for developers who are addressing edge AI.

Empowering edge AI developers

As a semiconductor supplier, ST helps developers tackle these hardware and software challenges by providing many portfolios of different hardware devices and software development tools to address the different requirements of a growing number of AI-enabled applications.

For instance, the STM32 has become the leading solution in terms of number of projects that have been submitted in the MLPerf Tiny benchmark, with 73% of all the projects submitted based on STM32.

“I think this is related to three main factors,” says Maravita. “The first point is our leadership with our general purpose MCU STM32 for industrial and consumer application, our commitment as a contributor in edge AI benchmarks since the inceptions of them like the MLPerf Tiny benchmark, and finally, our online platform, the STM32 AI Developer Cloud, which helped customers and developers to easily test their models on a wide variety of STM32 boards with our online tools. We believe all this has encouraged widespread AI innovation on the STM32.”

Dealing with AI algorithms, however, requires a different skillset—software development. “The embedded software engineer needs to be focused on the implementation on edge AI and how to integrate it in the overall system,” says Maravita. “So, he may need to have a firmware project example to start with and to customize it for the specific application use case. The machine learning engineer, or AI engineer, or data scientist, will focus mainly on the development of the machine learning model. He not only needs to focus on the data set and the machine learning model, but also on the optimization of it for the specific hardware device that will be selected. The hardware engineer needs a simple tool to test the AI algorithm given by the data scientist on different hardware platform or different part numbers and to find the best compromise in term of performance, power consumption, size, price and so on.”

In comes ST Edge AI Suite, an integrated set of software tools free-to-use with ST hardware. ST Edge AI Suite enables customers to jumpstart the design and deployment of billions of connected, autonomous things embedding AI locally by simplifying the development of customers’ AI solutions by exploiting ST’s broad range of hardware—general-purpose and automotive MCUs and MPUs, and smart sensors—and related tools for embedded AI optimizations. The software suite expands and integrates the multiple software tools, evaluation, and development kits introduced over the past 10 years, while leveraging ST’s existing AI ecosystem of ML frameworks and key partners, such as Nvidia and AWS.

Meanwhile, the following are a few examples of how the company is working with the industry to enable the development of edge AI solutions.

First, ST has cooperated with HPE Group to develop a virtual sensor that will help optimize and maintain EV motors using edge AI. The virtual sensor has an AI algorithm that works on ST’s latest automotive MCU, the Stellar family, and takes external data from sensors to extrapolate and estimate the internal rotor temperature of the motor—which is obviously not physically accessible. Apart from this, the Stellar MCU may run additional AI algorithms for predictive maintenance to identify anomalies that can be present in the systems, such as buffer, mechanical, and electrical.

For the personal computing and laptop market, ST has worked with HP to develop a smart sensor technology to optimize the power monitoring of laptops by avoiding overheating and battery drain. Through ST’s six-axis IMU MEMS sensors, engineers were able to collect a comprehensive set of data from the IMU sensors in many different conditions to develop and train AI models that recognize different user activities based on device and user motion.

Another example is an STM32 MCU-powered washing machine that runs an AI algorithm, which uses the information about the status of the current of the motor to estimate the weight of the clothes inside the washing machine in all the different operations. Estimating with higher accuracy the weight of clothes will create a program driving the motor using exactly the current needed, and decrease the overall usage of water and detergent, data receiver, and the overall power. This AI-powered washing machine can save between 15% and 40% energy and water for a washing cycle.

Device development strategies

More and more manufacturers are developing edge AI, and in particular, endpoint AI applications, some of which requiring long-term power for always-on devices. To ensure efficient, low-power, and intelligent operations, Maravita mentions two key strategies to help customers develop such devices.

“The first strategy is to move some of the processing of the AI algorithms from the cloud to the edge,” he explains. “This is why we are focusing on edge AI solutions. By moving the processing locally, it will have a tremendous impact on saving the power consumption of the system, increasing responsiveness, increasing security, and decreasing the overall cost of the solution.”

The second strategy will be to select devices with integrated AI accelerators and that support SW tools.

“We mentioned before the NPU for the STM32 family, but we also have AI accelerators like the ISPU inside the MEMS. You can imagine that, for example, if you move your AI algorithm from an application processor to an MCU, you are already saving a lot of power. And if you can move, again, your model from a MCU to a MEMS sensor, you can again decrease the overall power, moving from milliampere order to the microampere order. So, you can have your AI algorithm always working inside the MEMS sensor at extremely low power consumptions, keeping the overall system in shutdown, and waking up only when needed.”

Sustaining innovation

Today, ST is recognized as one of the key leaders for edge AI worldwide, thanks to the company’s commitment to this sector over the past 10 years.

“We can see three dimensions of innovation. The first one is about hardware devices with the integration of hardware accelerators. The second is about software tools—we have seen specific AI libraries, the ST Edge AI Suite, and compiler specific for AI,” says Maravita. “The third dimension is about innovating by creating reference projects and proof of concepts for many use cases for our customers, helping them to innovate their products with new ideas using AI.”The global edge AI market is forecast to reach $107.47 billion by 2029, growing at a compound annual growth rate (CAGR) of 31.7% from 2022, according to Fortune Business Insights. Mainly driving this strong growth is the increasing application of AI in sectors including automotive, healthcare, personal computing, and industrial automation.

Indeed, AI has become indispensable in building a future connected world, where there are billions of things that are more secure, more connected, and more intelligent.

“We call this the Cloud Connected Intelligent Edge,” says Matteo Maravita, Senior Manager, AI Competence Center and Smartphone Competence Center, Asia Pacific Region, STMicroelectronics. “We believe that these devices will be much more autonomous and more connected to the cloud, thereby increasing not only the creation but also the local processing of data. All these devices cover several areas that we find in our daily lives.”

They can be found at homes, in factories, in workplaces, in cities and buildings, and in mobility solutions. “When we look at this wide variety of devices, and if we think about the related business model, we can understand how edge AI is going to be impacting our future a lot,” says Maravita.

Despite the continuous advancements in AI and machine learning (ML) technologies to enhance applications in many sectors, designers continue to face challenges—mostly hardware—when it comes to edge AI development.

“Basically, they want to implement AI, but they still need to look at the overall performance of the application,” says Maravita, He notes that designers need to ensure that the hardware they are going to use has a specific feature for security, for example, and to keep power consumption low.

Meanwhile, developers are also facing software challenges. For instance, they need to implement a learning model inside a machine. While they need to create, train, and deploy the model inside the final products, they also need to maintain this model. And this is a huge challenge for developers who are addressing edge AI.

Empowering edge AI developers

As a semiconductor supplier, ST helps developers tackle these hardware and software challenges by providing many portfolios of different hardware devices and software development tools to address the different requirements of a growing number of AI-enabled applications.

For instance, the STM32 has become the leading solution in terms of number of projects that have been submitted in the MLPerf Tiny benchmark, with 73% of all the projects submitted based on STM32.

“I think this is related to three main factors,” says Maravita. “The first point is our leadership with our general purpose MCU STM32 for industrial and consumer application, our commitment as a contributor in edge AI benchmarks since the inceptions of them like the MLPerf Tiny benchmark, and finally, our online platform, the STM32 AI Developer Cloud, which helped customers and developers to easily test their models on a wide variety of STM32 boards with our online tools. We believe all this has encouraged widespread AI innovation on the STM32.”

Dealing with AI algorithms, however, requires a different skillset—software development. “The embedded software engineer needs to be focused on the implementation on edge AI and how to integrate it in the overall system,” says Maravita. “So, he may need to have a firmware project example to start with and to customize it for the specific application use case. The machine learning engineer, or AI engineer, or data scientist, will focus mainly on the development of the machine learning model. He not only needs to focus on the data set and the machine learning model, but also on the optimization of it for the specific hardware device that will be selected. The hardware engineer needs a simple tool to test the AI algorithm given by the data scientist on different hardware platform or different part numbers and to find the best compromise in term of performance, power consumption, size, price and so on.”

In comes ST Edge AI Suite, an integrated set of software tools free-to-use with ST hardware. ST Edge AI Suite enables customers to jumpstart the design and deployment of billions of connected, autonomous things embedding AI locally by simplifying the development of customers’ AI solutions by exploiting ST’s broad range of hardware—general-purpose and automotive MCUs and MPUs, and smart sensors—and related tools for embedded AI optimizations. The software suite expands and integrates the multiple software tools, evaluation, and development kits introduced over the past 10 years, while leveraging ST’s existing AI ecosystem of ML frameworks and key partners, such as Nvidia and AWS.

Meanwhile, the following are a few examples of how the company is working with the industry to enable the development of edge AI solutions.

First, ST has cooperated with HPE Group to develop a virtual sensor that will help optimize and maintain EV motors using edge AI. The virtual sensor has an AI algorithm that works on ST’s latest automotive MCU, the Stellar family, and takes external data from sensors to extrapolate and estimate the internal rotor temperature of the motor—which is obviously not physically accessible. Apart from this, the Stellar MCU may run additional AI algorithms for predictive maintenance to identify anomalies that can be present in the systems, such as buffer, mechanical, and electrical.

For the personal computing and laptop market, ST has worked with HP to develop a smart sensor technology to optimize the power monitoring of laptops by avoiding overheating and battery drain. Through ST’s six-axis IMU MEMS sensors, engineers were able to collect a comprehensive set of data from the IMU sensors in many different conditions to develop and train AI models that recognize different user activities based on device and user motion.

example is an STM32 MCU-powered washing machine that runs an AI algorithm, which uses the information about the status of the current of the motor to estimate the weight of the clothes inside the washing machine in all the different operations. Estimating with higher accuracy the weight of clothes will create a program driving the motor using exactly the current needed, and decrease the overall usage of water and detergent, data receiver, and the overall power. This AI-powered washing machine can save between 15% and 40% energy and water for a washing cycle.

Device development strategies

More and more manufacturers are developing edge AI, and in particular, endpoint AI applications, some of which requiring long-term power for always-on devices. To ensure efficient, low-power, and intelligent operations, Maravita mentions two key strategies to help customers develop such devices.

“The first strategy is to move some of the processing of the AI algorithms from the cloud to the edge,” he explains. “This is why we are focusing on edge AI solutions. By moving the processing locally, it will have a tremendous impact on saving the power consumption of the system, increasing responsiveness, increasing security, and decreasing the overall cost of the solution.”

The second strategy will be to select devices with integrated AI accelerators and that support SW tools.

“We mentioned before the NPU for the STM32 family, but we also have AI accelerators like the ISPU inside the MEMS. You can imagine that, for example, if you move your AI algorithm from an application processor to an MCU, you are already saving a lot of power. And if you can move, again, your model from a MCU to a MEMS sensor, you can again decrease the overall power, moving from milliampere order to the microampere order. So, you can have your AI algorithm always working inside the MEMS sensor at extremely low power consumptions, keeping the overall system in shutdown, and waking up only when needed.”

Sustaining innovation

Today, ST is recognized as one of the key leaders for edge AI worldwide, thanks to the company’s commitment to this sector over the past 10 years.

“We can see three dimensions of innovation. The first one is about hardware devices with the integration of hardware accelerators. The second is about software tools—we have seen specific AI libraries, the ST Edge AI Suite, and compiler specific for AI,” says Maravita. “The third dimension is about innovating by creating reference projects and proof of concepts for many use cases for our customers, helping them to innovate their products with new ideas using AI.”

By: DocMemory
Copyright © 2023 CST, Inc. All Rights Reserved