Kaishita: a student-lead approach to afordable Vending Machines

---

Project Overview

On the mission to develop affordable, small, and customizable vending machines for everyone.

About Kaishita

Kaishita is a student-led project born at USP São Carlos, Brazil, aiming to revolutionize automated sales systems with compact and intelligent vending machines.

---

Goals

• Deliver automated sales solutions that fit any environment.
• Create a purchasing process as simple as grabbing the product (secure and theft-proof).
• Offer diverse payment methods like Pix and credit/debit cards.
• Provide real-time feedback to vendors, including stock management and AI-based pricing suggestions.

---

How It Works

The Kaishita vending machine integrates weight sensors, MQTT communication, and a centralized server to automate and analyze purchases.

Action Flow

1. User Interaction:
   The user selects the desired product and quantity to withdraw from the vending machine.
2. Server Logic:
   The server receives the request, monitors the event, and determines the state of the purchase:
   • Candies Taken (Successful)
   • Candies Not Taken (Unsuccessful)
3. Weight Verification:
   The ESP32 weight system ensures the correct amount of product has been taken.
4. Data Analysis:
   Every successful purchase is logged in a CSV file and automatically analyzed for trends.

---

Table of Contents

1. Project Overview
2. Goals
3. How It Works
4. Project Structure
5. Key Features
6. Setup Instructions
   • ESP32 Firmware
   • Analysis Script Setup
7. Example CSV File Format
8. Results
9. Project Images
   • Website UI
   • Serial Monitor Output
   • Project Setup
10. Contribute to the Project
11. Credits

---

Project Structure

Kaishita/
├── Hardware_Firmware/      # ESP32 firmware code for the vending machine
│   ├── components/         # External libraries and components
│   │   ├── arduino/        # Arduino libraries for ESP-IDF compatibility
│   │       ├── pubsubclient/   # MQTT library
│   │   ├── WiFiManager/    # Wi-Fi management library
│   ├── include/            # Header files for MQTT and weight logic
│   ├── main/               # Main source code
│   │   ├── 3Cell.cpp       # Weight scale logic (HX711 integration)
│   │   ├── mqtt.cpp        # MQTT communication
│   │   ├── main.cpp        # Firmware entry point
│   ├── CMakeLists.txt      # Build configuration
│
├── Website/                # Server code (handles vending machine communication)
│
├── Analysis/               # Automated purchase data analysis
│   ├── analise_compras.py  # Script for trend analysis
│   ├── requirements.txt    # Python dependencies
│   └── README.md           # Instructions for running analysis
│
└── README.md               # Main project documentation

---

Key Features

1. IoT Weight System

• Monitors the weight of products using HX711 load cells connected to an ESP32-S3.
• Publishes weight data to a server via MQTT for purchase verification.

2. Server Integration

• Handles user requests, verifies events, and tracks product states:
  • Product Taken
  • Product Not Taken

3. Data Analysis

• Logs successful purchases into a CSV file.
• Automatically generates:
  • Trend graphs: Showing product purchases over time.
  • Purchase probabilities: Calculating likelihood of sales at specific times (morning, afternoon, evening).

---

Setup Instructions

ESP32 Firmware

1. Clone the repository:

   git clone https://github.com/GUUDIN/Kaishita.git
   cd Kaishita/Hardware_Firmware/src

2. Install Arduino Component:

   git submodule add https://github.com/espressif/arduino-esp32.git components/arduino
   cd components/arduino
   git submodule update --init --recursive

3. Add Required Libraries:

   idf.py add-dependency "tzapu/WiFiManager"
   idf.py add-dependency "knolleary/pubsubclient"

4. Configure Wi-Fi and MQTT Broker in main/mqtt.cpp:

   const char* mqtt_server = "YOUR_MQTT_BROKER_ADDRESS";
   const int mqtt_port = YOUR_MQTT_BROKER_PORT;

5. Build and Flash Firmware:

   idf.py build
   idf.py flash

---

Analysis Script Setup

1. Navigate to the analysis folder:

   cd Kaishita/analysis

2. Install Dependencies in a Virtual Environment:

   python3 -m venv venv
   source venv/bin/activate   # Linux/MacOS
   pip install -r requirements.txt

3. Run the Analysis:

   python analise_compras.py

4. Provide the CSV file path when prompted.

---

Example CSV File Format

data	hora	produto	quantidade
2024-01-01	8	KitKat	10
2024-01-01	14	Mentos	5
2024-01-01	20	Bala	7

---

Results

1. Trend Graphs: Visualize total quantities of products sold at different times.
2. Purchase Probabilities: Calculate product popularity in the morning, afternoon, and evening.

---

See the current state of the project!

Website UI

Click to view Website UI

Serial Monitor Output

Click to view Serial Monitor Output

Project Setup

Click to view Project Setup

---

Contribute to the Project

If you want to collaborate:

1. Fork the repository.
2. Grab an issue and start coding!
3. Submit a pull request.

---

Credits

• ESP32 Firmware Development: Pedro Gudin & Mateus Messias
• Analysis and Server Integration: Mateus Messias & André Messias
• Documentation: Pedro Gudin & Mateus Messias & João Breches

---