Robotics Solution for Agriculture: Automated and IoT-enabled Tomato Picking and Packing

Abstract

This paper presents a novel robotics solution for the food industry, focusing on automated and IoT-enabled tomato picking and packing. The work includes a pilot execution of a low-cost implementation of Internet of Things (IoT) and Machine to Machine (M2M) communication for the postharvest stage, specifically targeting tomato picking and primary packing using a computer vision-based robotic arm. The vision system employs You Only Look Once (YOLO) model weights to accurately detect and localize tomatoes. The proposed IoT development enables users to send commands specifying the desired number of tomatoes to be packed, while also providing easy control and monitoring capabilities for workers. The command transmission occurs via a local network that incorporates a collaborative robot UR5, a Zivid Two 3D industrial camera connected to a laptop, and a Raspberry Pi microcomputer. This work addresses the crucial aspects of food safety and labor shortage in small and medium-sized enterprises (SMEs), while also paving the way for automated tomato processing to ensure the high quality of fresh, perishable tomatoes.