An Innovative Maker has developed an AI-powered robotic system designed to flip tofu while frying, reducing human exposure to hot splashing oil. The system integrates computer vision, robotics, and motion control to automate this delicate cooking task.
The robotic arm is mounted upside down on a gantry system for better reach across the frying pan. It uses two pairs of chopsticks, inspired by claw machine mechanisms, to grasp and flip tofu. Motion is controlled by stepper motors and servo motors, with a RAMPS 1.4 board on an Arduino Mega handling movement and a PCA9685 board managing servo control. A PID controller ensures smooth and stable operation by reducing jittering.
For tofu detection, the system employs YOLOv8-based AI, trained on 1,675 tofu images using the Roboflow platform. Once detected, the system calculates the tofu’s coordinates relative to the camera and guides the robotic arm accordingly.
Powered by a 12V power supply, the system currently flips one tofu piece at a time due to dataset limitations. Future improvements aim to expand the dataset for multi-tofu flipping and enhance motion sequences to place fried tofu on a plate.
This project showcases a creative blend of deep learning, robotics, and automation, making frying safer and more efficient.
