Version: FILS English

🔐 Smart Door Lock with Face Recognition

A secure, AI-powered door lock system using ESP32-CAM and Raspberry Pi Pico 2.

info

Author: Matei-Alexandru Tiță
GitHub Project Link: https://github.com/UPB-PMRust-Students/project-MateiTita

🧾 Description

This project implements a smart door lock system that uses face recognition for access control.
The system relies on an ESP32-CAM to detect and recognize faces and streams video live over Wi-Fi.
When a known face is detected, it sends a signal to a Raspberry Pi Pico 2, written in Rust, which activates a servo motor to unlock the door and triggers visual (LED) and audio (buzzer) feedback.

All access events are logged in real time to a connected laptop via USB.

💡 Motivation

The goal of this project is to explore a practical application of embedded systems and machine vision, combining:

Offline, low-cost AI via ESP32
Embedded Rust programming
Real-time feedback and logging
A functional and extendable prototype for smart home security

🧠 Architecture

System Flow

ESP32-CAM
- Captures live video
- Performs face detection and recognition
- Streams to browser via local IP
- Sends "GRANT" or "DENY" to Pico over UART
Raspberry Pi Pico 2
- Waits for serial commands
- Controls:
  - Servo for door lock
  - LEDs (Green = granted, Red = denied)
  - Buzzer for audio feedback
- Sends event logs to laptop via USB
Laptop
- Runs a Python script to receive and timestamp events

Hardware image

Scheme

🗓️ Development Log

Week 1 – Initial Planning

I set up and powered the Raspberry Pi Pico 2 and ESP32-CAM, tested the ESP32-CAM by flashing the CameraWebServer example and viewing the live video stream on my laptop, and verified the Pico environment by writing a simple Rust program to blink an LED, confirming basic GPIO control and that both boards were functioning properly.

Week 2 – Core Setup

In Week 2, I replaced the initial UART-based communication with a Wi-Fi-based protocol for transmitting access control signals between the ESP32-CAM and the Raspberry Pi Pico 2. I configured the ESP32-CAM to run a lightweight HTTP server that, upon recognizing a known face, sends a GRANT or DENY signal as an HTTP GET request to the Pico's IP address on the local network.

Week 3 – Hardware Integration

Week 4 – Logging and Testing

🔩 Hardware

Components Used

Component	Purpose
ESP32-CAM	Face recognition and video streaming
Raspberry Pi Pico 2	Hardware control (Rust)
Servo Motor (SG90)	Lock/unlock mechanism
Red and Green LEDs	Visual status indication
Buzzer	Audio feedback
Breadboard + Wires	Connections and prototyping
Laptop (USB)	Real-time log receiver

Bill of Materials

Component	Qty	Price (RON)	Total (RON)
ESP32-CAM	1	50	50
Raspberry Pi Pico 2	1	35	35
SG90 Servo	1	20	20
LEDs + Buzzer	4	5	5
Breadboard + Wires	1	15	15
Total			125

🔐 Smart Door Lock with Face Recognition

🧾 Description

💡 Motivation