Automatic Fruit Detection Based on Robotic Vision

Automatic fruit detection has greatly reduced labor costs and crop damage rates, contributing to the progress of agricultural modernization. It involves real-time assessment of the surrounding environment and recognition of target fruit categories. However, fruit detection faces challenges due to the similarities in shape, color, and other features among different types of fruits, as well as the impact of environmental factors such as inadequate lighting, occlusion, and background variations on recognition accuracy. This project focuses on the simulation of the robotic fruit harvesting process, which includes environmental image acquisition, image processing, and robot control. The robot utilizes its integrated camera to capture real-time images of the surrounding environment. These images are then processed on a computer, employing color recognition based on traditional algorithms. The images are segmented based on the identified color regions and subjected to object detection using YOLO v5. By discerning the presence of the target objects, the results are transmitted to the robot, which responds accordingly. The real-time performance is enhanced by filtering and narrowing down the target regions, reducing the time required for object recognition.