SiliAI: AI-Driven Robotic System for Efficient Chili Pepper Sorting Using YOLOv8

Addressing the increasing demand for chili peppers, this study develops the SiliAI system, an advanced AI-driven robotic arm tailored for sorting peppers into red, green, and rotten categories. The system integrates YOLOv8, a state-of-the-art object detection algorithm, with a 6-degree-of-freedom (6DOF) robotic arm and a 2-finger angular gripper, which collectively enhance sorting efficiency and accuracy. The SiliAI system achieves a promising peak F1 score of 93% and perfect precision 100% across all classes at specific confidence thresholds, demonstrating high reliability in classification. Precision-Recall Curves reveal impressive precision rates for GREEN (99.5%), RED (97.0%), and ROTTEN (91.3%) categories, with a macro-average precision of 95.9% at an Intersection over Union (IoU) of 0.5. These metrics underscore the system's ability to accurately sort peppers based on ripeness. Training metrics indicate consistent performance, supported by a balanced dataset and effective tool integration, including Arduino IDE, Roboflow, and Python. The SiliAI system not only promotes operational efficiency but also aligns with Sustainable Development Goals (SDGs) related to food security. Future plans involve expanding the dataset and refining the gripper to further enhance field performance. This research highlights the potential of AI-driven robotic systems to significantly improve productivity and reduce labor costs in agricultural sorting tasks.