What if building AI for tiny devices no longer needed multiple tools and setup steps? A platform connects data collection, model integration, and deployment.
Nuvoton Technology has introduced NuML Studio, a new graphical development platform designed to simplify machine learning deployment on microcontrollers. The software combines data collection, AI model integration, and firmware generation into a single workflow, reducing the setup complexity that often slows Endpoint AI development.
The platform is aimed at developers building edge AI applications on Nuvoton MCUs, particularly those using the NuMicro M55M1 family. Instead of requiring multiple development tools and manual configuration steps, NuML Studio provides an integrated environment that can move projects from sensor data capture to deployable firmware with minimal setup.
One of the main focuses of the tool is accessibility. NuML Studio runs on Windows and comes as a ready-to-use package, eliminating the need to install Python environments or additional software libraries. Developers can immediately begin creating projects for data collection, AI deployment, or combined workflows without spending time configuring dependencies.
The software also includes built-in support for collecting data from several sensor types through the NuMaker-M55M1 platform. Supported inputs include 3-axis motion sensors, 16kHz audio capture, and image collection. Once captured, raw data can be automatically converted into commonly used formats such as CSV, WAV, and JPG files, simplifying the preparation process for machine learning training pipelines.
NuML Studio further integrates with cloud-based machine learning platforms through API support, allowing developers to upload datasets directly for model training and evaluation.
On the deployment side, the platform supports TensorFlow Lite Micro and quantized AI models. It can automatically generate firmware projects for Keil MDK and VS Code CMSIS environments, supporting applications such as image classification, object detection, and keyword spotting.
For hardware acceleration, the software includes optimized libraries for chips equipped with the Arm Ethos-U55 NPU. This includes the NuMicro M55M1 series, where the NPU is used to improve inference performance and reduce processing overhead for edge AI workloads.
By consolidating data collection, model integration, and firmware generation into a single interface, NuML Studio is intended to shorten development cycles and lower the technical barriers associated with embedded AI projects.
Click here for the original announcement.
