Nuvoton Unveils Endpoint AI Platform for Machine Learning

Print Friendly, PDF & Email

Hsinchu, Taiwan, January 5, 2024 – Nuvoton has announced the Endpoint AI Platform designed to accelerate the development of fully-featured microcontroller (MCU) AI products. These solutions are enabled by Nuvoton’s  new MCU and MPU silicon, including the NuMicro M55M1 equipped with Ethos U55 NPU, NuMicro MA35D1, and NuMicro M467 series. These MCUs are a valuable addition to the modern AI-centric computing toolkit and demonstrate how Nuvoton continues to work with Arm and other companies to develop an Endpoint AI Ecosystem.

Development on these platforms is made easy by Nuvoton’s NuEdgeWise: a well-rounded, simple-to-adopt tool for machine learning (ML) development, which is nonetheless suitable for cutting-edge tasks. Together, this powerful core hardware, combined with unique rich development tools, cements Nuvoton’s reputation as a leading microcontroller platform provider.

These new single-chip-based platforms are ideal for applications including smart home appliances and security, smart city services, industry, agriculture, entertainment, environmental protection, education, highly accurate voice-control tasks, and sports, health, and fitness.

The NuMicro M55M1 series microcontroller is targeted at machine learning applications, aided by its Ethos-U55 NPU (Neural Processing Unit) and on-device AI features suitable for embedded applications. This MCU lets the system watch for events – based on image sensor, microphone, and sensors – while in low-power mode, without waking up the central processor. The M55M1 MCU includes an ML model protection mechanism that enhances security by safeguarding ML intellectual property against potential malicious hacking attempts. These are some of the first processors to support Arm Helium technology, which provides a significant performance boost for machine learning (ML) and digital signal processing (DSP) applications in small, low-power embedded systems.

The MA35D1 series is a heterogeneous multi-core microprocessor for high-end Edge IIoT Gateway, based on a dual-core 64-bit Arm Cortex-A35 core at 800 MHz and a 180 MHz Arm Cortex-M4. These high-performance cores facilitate Tiny AI/ML edge computing.

The M467 series is a 32-bit microcontroller based on the Arm Cortex-M4F core with a built-in DSP instruction set and single precision floating point unit (FPU). It is ideal for a wide range of applications: smart home appliances, IoT gateways, industrial control, telecommunications, and data centers.

In IoT tasks, the M467 can be enhanced with a rich set of connectivity, I/O, and security peripherals, from Ethernet 10/100 MAC to hardware encryption, decryption, and key storage. With the M467’s broad built-in I/O support, users can choose only the precise hardware extensions they need for their particular applications. The M467 also supports HyperRAM. In AI/ML applications, the 64MB of HyperRAM provides the flexibility to handle different ML models with varying memory size or density requirements. HyperRAM also offers power-saving, suitability for available bandwidth, ease of use, and flexible expansion of memory options.

Fully-featured development boards are available for all of the above hardware applications. These are supported by Nuvoton’s deep development tools, development environment, and enthusiastic support. For example, AI application development with NuMaker MA35D1 not only enables efficient machine learning projects, such as image classification but also presents analysis to the user intuitively via the Human Machine Interface (HMI). Meanwhile, the NuMaker-IoT-M467 development board is specifically designed for IoT applications of the M467 MCU.

Nuvoton’s NuEdgeWise IDE (Integrated Development Environment) is a machine-learning tool designed for TinyML development. The IDE supports the four key stages of ML application development: labeling, training, validation, and testing. NuEdgeWise leverages the popular Jupyter Notebook platform, allowing developers to train and deploy models on Nuvoton chips using TensorFlow Lite. This makes TinyML applications more accessible and easier to implement.