Major project
Electronics, Semiconductors & Telecom
TinyML on a microcontroller
Train a small model and run it on an ESP32 or Raspberry Pi Pico, measuring speed, size and power, with updates over the air.
The brief
Smart devices need AI that runs on a few hundred kilobytes and a battery. Train a small model (keyword spotting, gesture or anomaly detection), shrink it, deploy it on a microcontroller, and measure accuracy, latency, memory and power; then add a safe way to update it. Good looks like: a model within the device's limits, measured trade-offs, and a working device demo.
What you will build
- Choose a task and collect or reuse sensor data
- Train a small model
- Quantise and convert it for the microcontroller
- Deploy and measure latency, memory and power
- Compare accuracy on the device with the original
- Design an over-the-air update
Free, open data to build it with
- LiteRT for microcontrollers ↗ Google AI Edge
- Edge Impulse ↗ Edge Impulse
Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.