With the introduction of custom Neuton models and the Axon NPU, Nordic Semiconductor has significantly expanded the on-device AI capabilities of its product lineup. Custom Neuton models are designed for CPU-run edge AI and are up to 10x smaller, achieving up to 10x higher efficiency than standard LiteRT models running on the CPU. Neuton models are suitable for simpler edge AI applications, like interpreting data from time-series sensors like accelerometers.

For more advanced use cases, we’ve also introduced the Axon NPU, our dedicated AI accelerator core for standard LiteRT models. It offers up to 15x faster inference and greater energy efficiency than running the same model on the CPU. The Axon can support use cases not viable for CPU-run edge AI, such as interpreting high-rate sensor data or sensor fusion, as well as audio classification, for example, for voice commands.

In this webinar, Robin from Nordic Semiconductor will walk you through the complete Nordic Edge AI ecosystem, including hardware, software, tools, and how to get started with edge AI in your nRF Connect SDK project.

Agenda:

• Intro to custom Neuton models and nRF54LM20B SoC with Axon NPU

• Walkthrough of the Nordic Edge AI Lab

  • Custom Neuton models for CPU-run edge AI

  • Text-to-wake-word for Axon NPU

  • Model builder for Axon NPU

• Intro to the Edge AI Add-On for nRF Connect SDK

  • Where to find and how to install the add-on

  • Overview of the nRF Edge AI API

• Live Q&A session where the experts will answer your questions