STMicroelectronics Deep Edge AI, a new force in artificial intelligence, emerges

1. Introduction to AI

AI (artificial intelligence) originated from a summer seminar held by Dartmouth College in 1956. At the conference, the term "artificial intelligence" was formally proposed for the first time. Technological breakthroughs in computing power have driven round after round of development of artificial intelligence. In recent years, with the increasing availability of big data, the third wave of artificial intelligence development has arrived. In 2015, artificial intelligence algorithms based on deep learning surpassed humans in image recognition accuracy in the ImageNet competition for the first time, and artificial intelligence has made great strides in its development. With breakthroughs in computer vision technology research, deep learning has achieved great success in different research fields such as speech recognition and natural language processing. Now, artificial intelligence has shown great potential in all aspects of life.

In light of the development stage of artificial intelligence technology, some main concepts are roughly explained as follows.

AI: Any technology that enables a computer brain to simulate human behavior.

Machine Learning: A subset of Artificial Intelligence (AI). Algorithms and methods that continuously improve by learning from data.

Deep learning: A subset of machine learning (ML). Learning algorithms that extract valuable information from large amounts of data using a multi-layered structure that mimics the neural network of the human brain.

2. STMicroelectronics Deep Edge AI, a new force in artificial intelligence

At present, due to the demand for computing power, artificial intelligence technology is mainly used in cloud scenarios. Due to limitations such as data transmission delays, cloud-based solutions may not meet the needs of some users for data security, system responsiveness, privacy, and local node power consumption. In centralized artificial intelligence solutions, embedded devices (smart speakers, wearable devices, etc.) usually rely on cloud servers to achieve artificial intelligence capabilities, while in Deep Edge AI solutions, embedded devices themselves can run artificial intelligence algorithms locally to achieve real-time environmental perception, human-computer interaction, decision-making control and other functions.

Moving the inference process to deep edge computing brings some advantages, such as system responsiveness, better privacy protection of user information (not all data needs to be transmitted to the cloud through multiple systems), and reduced connection costs and power consumption.

According to ABI research, global shipments of Deep Edge AI devices will reach 2.5 billion units by 2030. STMicroelectronics has noticed that there are more and more communities and ecosystems around Deep Edge AI technology, focusing on independent, low-power and cost-effective embedded solutions. As a major driver of this trend, STMicroelectronics has invested a lot of resources in AI to help developers quickly deploy AI applications on embedded systems based on microcontrollers/microprocessors (STM32 series) and sensors (MEMS, ToF...). STMicroelectronics provides a set of AI tools for the STM32 series and MEMS sensors with integrated machine learning cores (MLC) to speed up the development cycle and optimize trained AI models (STM32Cube.AI).

As a general technology, artificial intelligence has achieved remarkable achievements in many fields. We believe that more and more smart terminal devices will have a more direct and positive impact on human life.

3. Rapidly deploy AI applications through ST’s ecosystem

STMicroelectronics provides an ecosystem of hardware and software to enable quick and easy development of a variety of Deep Edge AI algorithms for sensors and microcontrollers.

Machine Learning in the MEMS Sensor Ecosystem helps designers leverage AI at the Edge for gesture, activity recognition, anomaly detection, and more through a decision tree classifier running on the sensor’s embedded engine called the Machine Learning Core (MLC).

Therefore, IoT solution developers can deploy any of our sensors (with embedded machine learning core) in a rapid prototyping environment to quickly develop ultra-low power applications using the UNICO-GUI tool.

With built-in low-power sensor design, advanced AI event detection, wake-up logic and real-time edge computing capabilities, the MLC in the sensor greatly reduces the amount of system data transmission and reduces the network processing burden.

If developers decide to develop a solution based on an in-sensor machine learning core, they will need a completely new approach to publish their application.

The starting point for creating any machine learning algorithm is the data and its definition of classes that describe the complex problem to be solved. You can follow five steps to create and run AI applications in sensors. UNICO-GUI is a graphical user interface that supports all five steps, including decision tree generation.

In order to facilitate developers to quickly deploy trained AI models to STM32, we have developed an easy-to-use and efficient tool - STM32Cube.AI (also known as X-CUBE-AI). X-CUBE-AI can analyze and convert trained neural networks into optimized C language code and automatically test them for STM32 targets. Of course, X-CUBE-AI is a very powerful tool, and more of its features will be introduced in subsequent articles.

To demonstrate how several different AI applications can run directly on the STM32 and accelerate the development, verification, and deployment process for STM32 embedded developers, STMicroelectronics provides a number of AI applications as references.

Developers can conduct secondary development based on these embedded AI application software packages and quickly deploy customized models.

More details will be introduced in subsequent articles.

AI development tools and embedded application software packages are summarized as follows

Embedded Software

Wherever there is STM32, there is Deep Edge AI.

All STM32 MCUs support the deployment of AI models. For MCUs with lower computing power, they support machine learning algorithms (ML). For MCUs with higher computing power, they also support neural network models (DL).

The list of evaluation boards that can run the application examples is summarized below.

Product Evaluation Tools

4. Want to know more details?

We will publish a series of articles detailing ST’s efforts in the field of Deep Edge AI.

Please let us know in the comments what you would like to know about STMicroelectronics AI and we will bring you more exciting content.