Waste Classification Using Machine Learning and SCARA Robotic Arm This project introduces an innovative solution to automate waste segregation using a SCARA Robot enhanced with computer vision and machine learning. The project used a pre-deployed AI model to classify the objects into organic or inorganic waste. At its core, the project utilises a Raspberry Pi paired with a USB camera to capture images of waste materials, this captured image is then processed with the help of inference SDK and the Roboflow waste classification API. Once classified, the robotic arm, controlled by an ESP32-C6 WROOM module, handles the waste, placing it into designated bins. This integration of robotics and artificial intelligence not only enhances automation but also contributes to effective waste management solutions.
The SCARA robot retains its compact and precise design, featuring NEMA 17 stepper motors powered by TMC2209 drivers for smooth operation, alongside SG90S servo motors for precise gripper control. The robust power system includes an LM2596-5 buck converter and an ADP7118AUJZ-3.3 LDO, ensuring the reliable operation of the ESP32-C6 and its peripherals. The Raspberry Pi handles image capture, and ML inferencing, leveraging the RoboFlow platform API for real-time inference. A Python-based control script on the Raspberry Pi seamlessly interfaces with the ESP32-C6 via serial communication to coordinate waste classification and robotic arm movements. This project exemplifies the potential of combining robotics and AI to solve real-world challenges. It provides a hands-on platform for exploring automation, embedded systems, and machine learning applications.