AI-Augmented Robotic System for Agriculture: Intelligent Packing Application

This paper introduces an improved robotic solution for the agriculture and food industry, applying Artificial Intelligence (AI) in tomato picking and packing processes. The work proposes a cost-effective integration of Internet of Things (IoT) and Machine-to-Machine (M2M) communication during post-harvest processing of tomatoes, facilitated by a computer vision-based robotic arm. The system leverages You Only Look Once (YOLO) model for accurate tomato identification, localization and relative size estimation. Our IoT implementation enables users to send commands specifying the desired tomato quantities and sizes for packing. The system also offers seamless control and monitoring capabilities for the maintenance personnel. Command transmission occurs through a local network incorporating a collaborative robot UR5 and Zivid Two industrial 3D camera, interconnected through a laptop and a Raspberry Pi microcomputer. The proposed system has achieved picking success rate of 0.98 and packing success rate of 0.95. This work addresses concerns related to food safety and labor shortages in small and medium enterprises (SMEs) by contributing to the automation of tomato processing and ensuring consistent quality of fresh, perishable tomatoes.