FROM HARVEST TO PACKAGE: AN AUTONOMOUS ROBOT FOR INTEGRATED TOMATO PICKING AND BAGGING

Design and Experimental Validation of a Vision-Based Robotic Framework for Strawberry Harvesting

The automation of fruit harvesting has become an important research topic in precision agriculture due to increasing labor shortages, rising production costs, and the need for improved harvesting efficiency. Among horticultural crops, strawberries present particular challenges for robotic harvesting because of their variability in size, shape, ripeness, and frequent occlusions caused by leaves and surrounding fruit. The objective of this work is to demonstrate the feasibility of a reproducible perception-to-manipulation framework for robotic strawberry harvesting based on commercially available hardware and established computer vision techniques, rather than to propose a novel object detection algorithm. The proposed system integrates a YOLOv3-based (You Only Look Once) object detector, monocular vision for fruit localization, and a Universal Robots UR5e collaborative manipulator. Strawberry coordinates estimated from monocular images are transformed into the robot reference frame and transmitted through the XML-RPC (Extensible Markup Language-Remote Procedure Call) protocol, enabling robot positioning. The system was experimentally validated in a controlled indoor environment under different artificial illumination conditions. The YOLOv3 detector achieved a mAP0.5:0.95 of 37.4%, a precision of 84.2%, a recall of 76.1%, and a latency of 6.5 ms per image (153.8 FPS). The experiments also demonstrated reliable communication between the perception and robotic manipulation modules, enabling the robotic arm to reach the estimated strawberry positions. The proposed framework provides a practical and low-cost solution for integrating deep-learning-based perception with robotic manipulation and establishes a solid basis for future work on localization accuracy, automated grasping, harvesting efficiency, and deployment in real agricultural environments.

Score-Based Harvesting Order Planning and False Detection Handling for Robotic Tomato Harvesting

Vision-Guided Agricultural Robot for Crop Detection Using Edge Computing on Embedded Hardware

Height-Based Stratification in Greenhouse Harvesting Robotics: A Review from Ground-Level to High-Wire Crops

Labor shortages and the push for higher greenhouse efficiency have accelerated interest in automated harvesting. However, the development of a universal harvesting robot has been constrained by large variations in crop architecture, especially plant height. In this review, a height-based stratification of greenhouse harvesting robots and transferable high-wire crop harvesters is presented, covering ground-level crops (<0.6 m, e.g., strawberry), medium-height crops (0.6–1.5 m, e.g., tomato), and high-wire crops (>1.5 m, e.g., trellised cucumber). For each height layer, key design features, technical progress, prototype performance, and common obstacles—including fruit occlusion, mechanical crop damage, unreliable operation, and high commercial costs—are analyzed. Future efforts should target intelligent perception, soft end-effectors, and height-specific solutions (swarm robotics for ground crops, modular hybrid designs for medium crops, infrastructure co-design for high-wire crops). By using plant height as the primary stratification criterion, a design-oriented framework is provided, distinct from conventional crop-type or mechanism-based categorizations.

AI-Driven Closed-Loop Agricultural Robotics for Precision Phenotyping and Autonomous Selective Harvesting

This paper addresses the persistent gap between robotic crop monitoring and autonomous harvesting by proposing an integrated, closed-loop agricultural system that unifies real-time AI-driven phenotyping with selective fruit harvesting in a single operational framework. We developed and field-evaluated AgriLoop-1, a unified robotic platform combining a multi-modal perception suite, a compliant end-effector for damage-aware fruit handling, and a co-designed AI software architecture. The perception pipeline leverages a YOLOv8-Seg model to perform simultaneous fruit detection, instance segmentation, ripeness classification, and yield estimation, while a dynamic Harvestability Score guide’s target prioritization and optimized picking sequences based on spatial, phenotypic, and reachability constraints. Field trials conducted in a commercial apple orchard demonstrated reliable real-world performance, achieving an operational success rate of 87.0% and an effective harvesting throughput of 122 fruits per hour over 15 hours of continuous deployment. The system maintained stable closed-loop operation under varying environmental conditions, confirming the robustness of the integrated perception–decision–action architecture. This study presents a field-validated robotic system that effectively closes the perception–action loop in agricultural robotics. By tightly integrating precision phenotyping with adaptive harvesting decisions, the proposed framework moves beyond conventional modular approaches and establishes a practical foundation for intelligent, data-driven crop management systems capable of addressing labor shortages and supporting sustainable agricultural production.

STEMbot: A Compliant Robot for Under-Canopy Plant Navigation

The scalability of organic agriculture is partially limited by the labor costs associated with monitoring for pests. While drones and rovers are well-suited for agricultural monitoring from above or next to plants, many pests live on the underside of leaves or on plant stems, making them detectable only after they have caused significant damage. To enable early pest detection we present STEMbot, a miniature climbing robot system designed for autonomous navigation under plant canopies. Unlike existing climbing platforms that lack on-board perception or are restricted to unbranched vertical trunks, STEMbot integrates a fully geometric PIN-SLAM pipeline with a semantic OcTree to achieve robust localization and mapping while climbing the plant. To plan STEMbot's motion we propose a manifold-constrained A* planner along with ray-tracing goal specification to enable branch-aware traversal and the inspection of occluded targets. We validate our system through hardware experiments, demonstrating reliable traversal of stems ranging from 7-33mm and autonomous navigation across four distinct plant specimens. Quantitative evaluations show that our system achieves high-fidelity geometric reconstructions with an average Chamfer distance of less than 1cm relative to an offline photogrammetry baseline, confirming that STEMbot maintains the globally consistent odometry needed for autonomous navigation.