A student team from Politecnico di Milano and Mälardalen University built an edge AI coffee machine on an ESP32-P4 microcontroller as part of a distributed software development course. The project features facial recognition (ESP-DL), voice command support (ESP-SR), and a touch-driven UI built with Qt for MCUs — all running fully offline on the device. Students split into UI/UX, backend, and AI sub-teams, tackling challenges like QML learning curves, resource constraints, and multi-model integration. They solved performance issues by manually assigning AI models to specific processor cores. The result showcases how face recognition, speech recognition, and adaptive drink recommendations can run simultaneously on constrained embedded hardware.
Table of contents
The Challenge: Build a Complete Edge AI Application on Resource‑Constrained HardwareMeet the Edge AI Coffee MachineMeet the Student TeamInspiring the Next GenerationSort: