Answers are grounded in this paper and cite their sources.
The increasing demand for agricultural products, along with rising production costs, has exacerbated labor shortages in the agricultural sector, particularly for crops that require manual harvesting, such as grapes, which makes minimizing transportation times critical for preserving produce quality. In this work, an autonomous robotic fleet for assisting in the grape harvesting process is designed and implemented, with the aim of improving efficiency while maintaining product quality. The fleet operates through four key states: autonomous navigation to an assigned operator, tracking of the operator as grapes are deposited into a harvesting box that is carried by a robot, replacement by another robot with an empty box, and autonomous navigation to an unloading point with the full box. The system integrates a GNSS RTK receiver and an IMU sensor for accurate localization and navigation within vineyard rows, thereby enabling reliable coordination between multiple robots in a confined agricultural environment. Real-time communication is performed via MQTT, which enables telemetry data exchange between the robots and a centralized mission control center that is responsible for fleet coordination and task allocation. Field experiments demonstrated successful execution of all behavioral states, with an average navigation accuracy within 0.3 meters and an obstacle avoidance success rate above 95%, despite the challenges posed by vineyard terrain and dynamic obstacles. Minor timing deviations were observed during robot replacement and navigation transitions, which were mainly because of dynamic obstacle interactions and path replanning events, highlighting areas for future improvement through optimized replacement distances. The presented fleet model is adaptable to diverse manual fruit-picking applications and offers a scalable solution for addressing labor shortages and optimizing harvesting operations.
No takes yet. Share an insight, caveat, or question.
Conejero et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: