The STM32N6 boasts a remarkable 600-fold enhancement in machine learning efficiency compared to a high-end STM32 microcontroller, setting a new benchmark in modern computing. | Credit score: STMicroelectronics
The latest advancements in microcontrollers (MCUs) promise to empower developers to integrate more artificial intelligence into edge devices such as robots. STMicroelectronics has unveiled a novel microcontroller (MCU) series, boasting its most powerful iteration yet, which incorporates accelerated machine learning capabilities.
“We are poised to initiate a significant metamorphosis at the very threshold.” According to Jean-Marc Chancel, President of the Microcontrollers, Digital ICs and RF Products Group at STMicroelectronics, this transformation involves the augmentation or supplementation of our clients’ workloads through AI-powered solutions. Currently, these fashion trends serve purposes equivalent to segmentation, classification, and recognition. As technology advances, these tools will inevitably be repurposed for novel applications, requiring further development.
Geneva, Switzerland-based STMicroelectronics announced that it employs over 50,000 staff members worldwide who are dedicated to developing cutting-edge semiconductor technologies. As the leading machine producer, we partner with more than 200,000 customers and thousands of collaborators to craft innovative solutions, methodologies, and ecosystems that address their unique challenges.
STM32N6 consists of new options
STMicroelectronics has designated a specific MCU as the primary platform to integrate its custom-built neural processing unit (NPU), dubbed, engineered to efficiently execute AI-related computations at the edge. According to claims, the MCU purportedly boasts the capability to deliver approximately 600 times greater machine-learning performance compared to a high-end STM32 MCU currently available.
The STM32N6 microcontroller features an 800 MHz ARM Cortex-M55 core, providing ample processing power, while its 4.2 MB of RAM ensures efficient handling of demanding data-intensive applications. STMicroelectronics has unveiled the Neural-ART Accelerator, a high-performance processor capable of operating at 1 GHz and delivering up to 600 GOPS (gigaoptions per second). This enhanced processing power enables real-time neural network inference for applications in laptops and audio functions.
The STM32N6 marks its debut in the MCU with a unique addition: a picture signal processor (PSP), empowering straightforward integration of cost-effective image sensors, according to the company. The Internet Service Provider (ISP) will be configured using a complimentary software programme that enables customisation of parameters mirroring those typically found in marketing campaigns, including exposure time, prominence, and lightness consistency.
According to El-Ouazzane, the STM32N6 will fully leverage the capabilities of our novel AI software ecosystem suite, marking its debut. “This milestone initiates a comprehensive expedition for AI hardware-enabled STM32, unlocking enhancements in functionality and product offerings through means previously inaccessible with other embedded processing solutions.”
STMicroelectronics has integrated a dedicated laptop vision pipeline featuring a MIPI CSI-2 interface, ensuring seamless compatibility with various cameras. The STM32N6 also boasts an H.264 hardware encoder and the NeoChrom Accelerator for enhanced graphics capabilities, positioning it as a suitable choice for feature-rich products.
Available in six distinct configurations spanning 169 to 264 pins and a range of pitches from 0.4 to 0.8 millimeters, this microcontroller boasts an impressive tolerance for ambient temperatures up to 125 degrees Celsius, rendering it suitable for use in extreme environments?
According to Tom Hackenberg, principal analyst in the Reminiscence and Computing Group at Yole Development, a widespread misconception exists that AI is solely an extensive knowledge hub, a power-hungry software.
He flatly denied the statement. “The current state of IoT edge functions is craving the insights and intelligence that AI-powered analytics can provide.” The STM32N6 exemplifies the convergence of power-efficient microcontrollers, AI-driven analytics, and computer vision capabilities, enabling seamless integration with vast amounts of sensor data to yield significant cost savings across modern equipment.
As AI and machine learning transform industries, STMicroelectronics’ microcontrollers (MCUs) are empowering the development of innovative applications. By leveraging the company’s cutting-edge expertise in embedded systems, developers can harness the power of AI to create intelligent devices that learn from data and adapt to new situations. With a wide range of MCU families offering varying levels of processing power, memory, and peripherals, STMicroelectronics provides the flexibility needed for diverse AI-based projects. Whether it’s developing smart home appliances or autonomous vehicles, MCUs are at the heart of creating AI-driven solutions that change lives.
STMicroelectronics announced a suite of software development tools designed specifically for the creation of edge-based machine learning functions, allowing users to “bring their own model” in formats such as TensorFlow Lite, Keras, and ONNX.
The launch of this innovative technology enables the creation of a comprehensive AI “ecosystem,” empowering developers to harness its capabilities for diverse applications with unprecedented speed and adaptability. Companies can now explore a diverse range of AI-powered solutions to accelerate their achievement and reduce time-to-market metrics.
The STM32N6 microcontroller boasts enhanced safety features, seamlessly meeting the latest industry standards to achieve SESIP Level 3 and PSA Level 3 certifications.
The STM32N6 microcontroller has been available for customer selection since October 2023 and is now capable of being supplied in large quantities. Companies such as LG Electronics are incorporating the STM32N6 microcontroller into their products.
The STM32N6 boasts exceptional AI performance while providing versatile flexibility within a compact silicon package ideal for embedded systems and wearables, according to Yehan Ahn, research and development process chief at LG’s Chief Technology Office. The inference pace, propelled by the trailblazing Neural-ART Accelerator, has surpassed our projections; meanwhile, ST’s developer-centric software toolkit enables effortless integration of our AI models with the microcontroller unit.
“Rigorously tested in our labs, the STM32N6 microcontroller from ST has impressed us with its exceptional neural processing capabilities and remarkable energy efficiency,” said Seiichi Kawano, Principal Researcher at Lenovo Research. “This microcontroller marks the beginning of a transformative era, significantly advancing our exploration of AI-powered innovations at the edge.”