This project uses a Raspberry Pi 5, a Pi Camera and a 2WD robot kit with an L298N motor driver to recognize three types of materials and move the robot along different paths depending on the detected material.
Instead of a pre-trained AI model, the system performs real-time material classification using computer vision techniques (OpenCV), based on:
- Brightness (V channel in HSV)
- Colour saturation (S channel in HSV)
- Edge density (Canny edge detection)
The system classifies materials as:
- Paper
- Plastic
- Metal
Shows the live camera image and the predicted material name on the screen.
Turns on a different LED for each material (for example: Green, Yellow, Red).
Makes the robot perform a different movement circuit:
- Material 1 (Paper) → Turn left 90° and then go forward 30 cm
- Material 2 (Plastic) → Go forward 30 cm
- Material 3 (Metal) → Turn right 90° and then go forward 30 cm
The robot uses a 2WD chassis with L298N motor driver and is powered from a 12V battery and a 12V→5V buck converter for the Raspberry Pi.
