Ezurio has launched a series of wireless computing modules designed to process data locally on factory equipment, enabling manufacturers to deploy predictive maintenance, industrial robotics, and asset tracking systems without relying on cloud connectivity.

By executing machine learning tasks directly on the device, the hardware allows industrial operators to monitor equipment health in real time and automate factory processes securely. The modules target applications in industrial wearables, predictive maintenance sensors, and robotics, providing ultra-low-power connectivity and offline decision-making capabilities for remote manufacturing environments.

The flagship BL54LM20B module features an integrated neural processing unit that delivers up to 15 times faster machine learning inferencing compared to standard processors. This edge computing capability allows sensors to analyse vibration or temperature data locally, identifying potential machine failures before they occur while reducing overall power consumption.

Built on Nordic Semiconductor architecture, the series also includes the BL54LM20A, which provides expanded memory and input/output options for complex industrial gateways. Both modules feature a 128-megahertz Arm Cortex-M33 processor, a RISC-V coprocessor, and support for Bluetooth Low Energy, Thread, and NFC protocols within an 11 by 8-millimetre footprint.

“The BL54LM20B shows exactly what that looks like in the field: a pre-certified module that gives developers up to 15x faster on-device inferencing without asking them to become NPU experts first,” said Sam Presley, technical product manager at Nordic Semiconductor.

The product line includes a USB-C adapter for integration into industrial computers and testing rigs. The modules share identical footprints and regulatory certifications, allowing manufacturers to upgrade existing hardware to AI-enabled versions without redesigning their circuit boards.