🔐 LockAI: IoT Multi-Touch Smart Lock

LockAI is an intelligent access control system built on the ESP32 platform. It uses multi-finger touch patterns to create a secure, low-cost smart lock. The system supports remote unlocking and monitoring via the Blynk mobile app.

This repository includes not only the firmware for the smart lock but also a complete data collection and machine learning pipeline. Python scripts are provided to capture touch patterns, convert them to a CSV dataset, and prepare them for training with services like Edge Impulse.

For a live demonstration and detailed project walkthrough, please visit our Project Website.

---

🚀 Key Features

• Multi-Touch Authentication: Requires a specific multi-finger touch combination for access.
• ML Ready: Includes firmware to log touch patterns and a Python converter to create ML-ready datasets (.csv).
• Remote Control: A Blynk-powered mobile app allows for remote unlocking, status monitoring, and configuration.
• Rule-Based & ML-Mimicking Logic: The current implementation uses a hard-coded, rule-based authentication logic derived from a collected dataset. This approach mimics the behavior of a trained ML model, as direct deployment of a .tflite model was not feasible due to library and dependency incompatibilities on the ESP32 platform.
• Comprehensive Hardware Support: Designed for the ESP32 with support for servo motors, capacitive touch sensors, and feedback LEDs.
• Complete Data Pipeline:
  • knock_logger.ino: Captures raw touch sequences and outputs them as JSON.
  • serial_data_logger.py: Reads JSON from the ESP32 and saves it to a .jsonl file.
  • jsonl_to_csv.py: Converts the .jsonl data into a flattened CSV, perfect for training in Edge Impulse.

---

🎯 Use Cases

• Personal Secure Storage: Lockers, drawers, or personal safes.
• Workspace & Lab Access: Secure access to shared equipment or restricted areas.
• Academic & Prototyping Projects: An excellent platform for learning about IoT, TinyML, and embedded systems.
• Low-Cost Surveillance: Log every interaction with a secure container, providing a digital trail of access attempts.

---

📦 Tech Stack

Layer	Technology / Platform
Microcontroller	ESP32 DevKit v1
Firmware	Arduino IDE (C/C++)
Sensors	TTP223 Capacitive Touch
Actuator	SG90 Servo Motor
Mobile App	Blynk IoT Platform
Data Logging	Python (PySerial)
ML Dataset Prep	Python (JSON, CSV)
ML Framework	Edge Impulse (for model training)
Website	React, Vite, GitHub Pages

---

🧠 System Architecture

The project operates in two main modes: Data Collection and Smart Lock Operation.

1. Data Collection Workflow

graph TD
    A[User Touch] --> B[ESP32 with knock_logger.ino]
    B --> C[Serial Port (JSON Output)]
    C --> D[Python: serial_data_logger.py]
    D --> E[knock_patterns.jsonl]
    E --> F[Python: jsonl_to_csv.py]
    F --> G[knock_patterns.csv]
    G --> H[ML Training (Edge Impulse)]

Loading

2. Smart Lock Operation Workflow

graph TD
    subgraph "Authentication"
        A[User Touch] --> B[ESP32 with Hard-Coded Logic];
    end
    subgraph "Remote Control"
        C[Blynk App] --> D[Blynk Cloud];
        D --> B;
    end
    B --> E{Access Granted/Denied?};
    E -- Granted --> F[Unlock Servo + Green LED];
    E -- Denied --> G[Red LED];

Loading

---

🗂️ Repository Structure

LockAI/
├── blynk_wifi/                 # Firmware for Blynk-only remote control
├── final_code-no_buzzer/       # Rule-based auth firmware without a buzzer
├── integrated_setup/           # Fully integrated firmware (touch, Blynk)
├── website/                    # Source code for the project's React-based website
├── knock_logger.ino            # ESP32 firmware for data collection
├── sample_code.ino             # Main smart lock firmware with ML and Blynk
├── serial_data_logger.py       # Python script to log serial data to a .jsonl file
├── jsonl_to_csv.py             # Python script to convert .jsonl to a .csv dataset
├── knock_patterns1.csv         # Sample dataset 1
├── knock_patterns2.csv         # Sample dataset 2
└── README.md

---

⚙️ Setup and Installation

1. Hardware Requirements

Component	Quantity	Purpose
ESP32 Dev Board	1	Main microcontroller
SG90 Servo Motor	1	Locking mechanism
TTP223 Capacitive Touch Sensor	3 (or more)	Multi-finger touch detection
LEDs (Red, Green)	1 each	Access granted/denied feedback
Breadboard & Jumper Wires	1	Prototyping and connections

2. Software & Configuration

1. Arduino IDE:

   • Install the ESP32 board definitions.
   • Install the following libraries from the Library Manager:
     • ESP32Servo
     • Blynk

2. Python:

   • Ensure you have Python 3 installed.
   • Install the pyserial library:

     pip install pyserial

3. Blynk:

   • Create a new project in the Blynk Console.
   • Create a new template and device.
   • Define datastreams for remote unlocking and status monitoring.
   • Find your Blynk Auth Token to use in the firmware.

---

🚀 How to Run the Project

Step 1: Data Collection (Optional, for custom ML models)

1. Flash knock_logger.ino onto your ESP32.
2. Connect the ESP32 to your computer via USB.
3. Modify serial_data_logger.py and set the SERIAL_PORT to your ESP32's port (e.g., COM3 on Windows, /dev/ttyUSB0 on Linux).
4. Run the logger:

   python serial_data_logger.py

5. Perform various touch patterns. The script will save them to knock_patterns.jsonl.
6. Once you have collected enough data, stop the logger (Ctrl+C) and run the converter:

   python jsonl_to_csv.py

   This will generate knock_patterns.csv, which you can upload to Edge Impulse for training.

Step 2: Deploying the Smart Lock

1. Open the integrated_setup/integrated_setup.ino file in the Arduino IDE.
2. Update the following credential placeholders with your own:
   • auth[]: Your Blynk Auth Token.
   • ssid[]: Your WiFi network name.
   • pass[]: Your WiFi password.
3. Flash the firmware to your ESP32. The lock is now operational.

---

🌱 Future Scope

• Multi-User Profiles: Support for multiple unique touch patterns for different users.
• Time-Based Access: Grant access only within specific time windows.
• Offline Mode: A fallback mechanism for when WiFi or cloud services are unavailable.
• OTA Updates: Over-the-air firmware updates for seamless improvements.
• Voice Commands: Integration with voice assistants for hands-free unlocking.

---

🤝 Contributors

This project was developed by:

• Varun Aditya
• Abhijay M. S.
• Shreyas Nayan Kamat
• Vaibhav V. S.

---

📄 License

This project is licensed under the MIT License. See the LICENSE file for more details.

---

🙌 Acknowledgements

• Edge Impulse for their powerful and accessible TinyML platform.
• Blynk for simplifying IoT mobile app development.
• The ISE Department at RVCE for providing the IDEA Lab platform and fostering innovation.