IoT-enabled robotic arm system capable of sorting multiple object categories with TFT display monitoring and web-based control dashboard.
Machine Learning • Computer Vision • Robotics • IoT • Embedded Systems
RoboSort Pro is an intelligent robotic automation system designed to automatically identify, classify, and sort objects based on multiple parameters including:
- Shape
- Color
- Size
- Weight
The system combines Machine Learning, Computer Vision, Weight Sensing, Robotic Manipulation, Embedded Systems, and Web-Based Monitoring into a single smart automation platform.
Unlike traditional sorting systems that rely on a single parameter, RoboSort Pro performs multi-parameter classification and sorts objects into 16 unique categories using a robotic arm.
The project was developed using a Raspberry Pi 4B, TensorFlow Lite, OpenCV, Flask, HX711 Load Cell Module, TFT Display, and a 6-DOF Robotic Arm.
- TensorFlow Lite image classification model
- Trained using Google Teachable Machine
- Edge AI inference on Raspberry Pi
- Real-time object recognition
The system identifies:
- Circle
- Rectangle
- Black
- White
- Small
- Large
using image classification and computer vision techniques.
The system uses:
- Load Cell
- HX711 Amplifier Module
to measure object weight.
Weight categories:
- Light
- Heavy
After classification:
- Object is picked automatically
- Target bin is determined
- Robotic arm places object into correct block
- Arm returns to home position
The system includes a web dashboard with role-based access.
Features:
- Login Authentication
- Start System
- Stop System
- Reset Statistics
- Live Monitoring
- Object Count Statistics
- User Monitoring
- Sorting Logs
Features:
- Read-Only Access
- Live Statistics
- Current Sorting Status
- System Monitoring
The TFT display provides:
- Object Prediction
- Confidence Score
- Weight Information
- Final Category
- Sorting Status
in real time.
The visual classifier predicts one of the following categories:
| ID | Category |
|---|---|
| 0 | Black Small Circle |
| 1 | Black Big Circle |
| 2 | White Small Circle |
| 3 | White Big Circle |
| 4 | Black Small Rectangle |
| 5 | Black Big Rectangle |
| 6 | White Small Rectangle |
| 7 | White Big Rectangle |
Visual classification is combined with weight classification to generate:
-
Black Small Circle Light
-
Black Small Circle Heavy
-
Black Big Circle Light
-
Black Big Circle Heavy
-
White Small Circle Light
-
White Small Circle Heavy
-
White Big Circle Light
-
White Big Circle Heavy
-
Black Small Rectangle Light
-
Black Small Rectangle Heavy
-
Black Big Rectangle Light
-
Black Big Rectangle Heavy
-
White Small Rectangle Light
-
White Small Rectangle Heavy
-
White Big Rectangle Light
-
White Big Rectangle Heavy
Total Categories: 16
┌─────────────────┐
│ Raspberry Pi 4B │
└────────┬────────┘
│
┌─────────────────────┼─────────────────────┐
│ │ │
▼ ▼ ▼
Pi Camera Module Load Cell + HX711 TFT Display
│ │
└───────────┬─────────┘
▼
TensorFlow Lite Model
│
▼
Classification Engine
│
▼
Sorting Decision Logic
│
┌───────────┴───────────┐
▼ ▼
Flask Dashboard PCA9685 Driver
│
▼
6-DOF Robotic Arm
│
▼
Sorted Bin
| Component | Quantity |
|---|---|
| Raspberry Pi 4B (8GB) | 1 |
| Raspberry Pi Camera Module | 1 |
| 6-DOF Robotic Arm | 1 |
| PCA9685 Servo Driver | 1 |
| MG996R Servo Motors | 3 |
| SG90 Servo Motors | 3 |
| Load Cell | 1 |
| HX711 Amplifier | 1 |
| TFT Display | 1 |
| Buck Converter | 1 |
| Lithium-Ion Batteries | 3 |
- Python
- TensorFlow Lite
- Google Teachable Machine
- OpenCV
- Picamera2
- NumPy
- Flask
- HTML
- CSS
- JavaScript
- Raspberry Pi GPIO
- PCA9685
- HX711
RoboSort-Pro/
│
├── src/
│ ├── app.py
│ ├── dashboard_data.py
│ ├── display_manager.py
│ ├── main_controller.py
│ ├── ml_classifier.py
│ ├── servo_control.py
│ └── weight_sensor.py
│
├── templates/
│ ├── login.html
│ ├── admin_dashboard.html
│ ├── viewer_dashboard.html
│ └── logs.html
│
├── model/
│ ├── model_unquant.tflite
│ └── labels.txt
│
├── dataset/
│
├── data/
│ ├── robot_data.json
│ └── sorting_logs.txt
│
├── testing/
│
├── images/
│
├── videos/
│
├── README.md
├── requirements.txt
└── .gitignore
- Object is placed on the platform.
- Camera captures object image.
- TensorFlow Lite model predicts visual category.
- Load Cell measures object weight.
- Weight category is assigned.
- Final category is generated.
- TFT display updates information.
- Dashboard statistics are updated.
- Sorting log is generated.
- Robotic arm picks object.
- Object is placed in assigned block.
- Arm returns to home position.
- Conveyor Belt Integration
- YOLO-Based Detection
- Cloud Dashboard
- Mobile Application
- MQTT Integration
- Industrial PLC Integration
- Voice Commands
- Predictive Analytics
- Advanced Deep Learning Models
This project demonstrates practical integration of:
- Artificial Intelligence
- Machine Learning
- Computer Vision
- Robotics
- Embedded Systems
- IoT
- Human-Machine Interfaces
- Industrial Automation
within a single intelligent automation platform.
Developed as a collaborative B.Tech (CSE - IoT) engineering project focused on smart automation and intelligent object sorting.
⭐ If you found this project useful, consider giving it a star.