Digital twin-driven system for efficient tomato harvesting in greenhouses

Abstract

Efficient and low-damage harvesting remains a major challenge in modern greenhouse tomato production, particularly in dense planting environments. To address limitations such as restricted camera views, occluded fruits, and complex fruiting patterns, our study presents a digital twin-driven system for intelligent tomato harvesting. Using a slidable depth camera mounted on the robot, we reconstruct a high-fidelity 3D digital twin of the greenhouse that accurately captures the spatial distribution and growth states of tomatoes. Based on this virtual environment, a learning-based framework is developed to optimize harvesting strategies, including robot positioning, arm trajectory planning, fruit selection priority, and adaptive operation modes. The proposed system integrates both a complete algorithmic workflow and a practical hardware platform. Experimental results show that our method significantly improves harvesting performance, reducing the average harvesting time by 34.95% (to 7.4 s per fruit), arm movement distance by 20.93%, and collision occurrences by 45.16%. While tailored for tomato harvesting, this framework demonstrates strong potential for generalization to other greenhouse crops in precision agriculture.