Our goal is to revolutionize sound and vibration sensing in environments where conventional rigid boards cannot be deployed. In small products or harsh environments, placing sensors in narrow gaps or on complex curved surfaces to acquire data has traditionally been an extremely difficult task.

Furthermore, connecting sensors by routing wired cables through thin piping poses significant challenges. 

As the demand for smart, real-time monitoring grows, there is a clear need for an unobtrusive and fully wireless solution to monitor physical activity and environmental conditions.

Starting with a focus on simplifying data acquisition in these challenging spaces, our prototype, the Ghost Edge AI Sticker, is an ultra-thin, flexible sensor node. As demonstrated in the attached images, the device utilizes a Flexible Printed Circuit (FPC) board, allowing it to act like a sticker that can be bent and attached directly to various curved surfaces (such as pipes) or installed in tight, narrow gaps.

The custom hardware is built around the nRF54L15 SoC and integrates a multi-sensor module featuring an accelerometer and a PDM sound sensor. To completely eliminate the need for cumbersome wiring, the device transmits data fully wirelessly via Bluetooth Low Energy (BLE). 

Furthermore, rather than relying solely on external processing, the on-board DSP core executes real-time Edge AI models, developed with Edge Impulse, to instantly infer and detect specific actions, vibrations, and voices. 

The firmware is built using Zephyr RTOS, enabling efficient multi-core applications that leverage Inter-Processor Communication (IPC) between the ARM Cortex-M33 and RISC-V cores.

Vibration and audio sensing is just the start! Using this versatile FPC platform, we are exploring broad use cases across various industries. By simply attaching the sticker, it becomes possible to understand exactly what is happening within a product or machine.

All of our custom KiCad hardware design files and diverse firmware projects are open-source, empowering the community to create the future of flexible Edge AI.

Reference: https://www.hackster.io/iotengineer22/ghost-edge-ai-sticker-908eb8