EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Autonomous Mango Harvesting Using YOLOv8 and ROS 2: A Simulation-Based Study

Osama Khalid, Muhammad Iftikhar

Food and Agriculture Communications · 11.06.2026 · 3 (1) · 23-33

DOI: 10.62887/fac.003.01.00113 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Food and Agriculture Communications, online erschienen am 11.06.2026. Verfasst von Osama Khalid und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Labor shortages, rising harvesting costs, and the need to reduce fruit damage are increasing interest in robotic harvesting. This study presents a complete simulation-based mango-harvesting pipeline that integrates a YOLOv8-Nano detector, ROS2 communication, and a UR10 six-degree-of-freedom robotic arm. A custom model was trained on 786 images to classify mangoes as ripe or raw. On the held-out test set, the model reached 97.1% mAP@0.5, 96.6% precision, 96.9% recall, and 30 frames per second. Only ripe fruit is selected for harvesting. To keep the sensing system simple and low-cost, the approach estimates relative depth from bounding-box width instead of using a dedicated depth sensor. ROS2 links the vision and control nodes with measured end-to-end latency below 150 ms. MATLAB simulation was then used to verify inverse kinematics, collision-aware motion, and a twist-and-pull detachment sequence for 15 target mangoes. The integrated simulation reported an 87% harvesting success rate and an average cycle time of 8.5 s per mango, equivalent to a theoretical cycle-rate throughput of about 424 mangoes per hour. These results show that a lightweight, camera-based architecture can support rapid and affordable development of robotic harvesters. However, the evaluation used only 30 test images and a simplified simulated canopy. Field trials are still required under changing light, wind, fruit occlusion, branch contact, and real gripper forces.

Zitieren

APA

Khalid, O., & Iftikhar, M. (2026). Autonomous Mango Harvesting Using YOLOv8 and ROS 2: A Simulation-Based Study. Food and Agriculture Communications, 3(1), 23-33. https://doi.org/10.62887/fac.003.01.00113

Vancouver

Khalid O, Iftikhar M. Autonomous Mango Harvesting Using YOLOv8 and ROS 2: A Simulation-Based Study. Food and Agriculture Communications. 2026;3(1):23-33. doi:10.62887/fac.003.01.00113

BibTeX
@article{Khalid2026,
  author = {Khalid, Osama and Iftikhar, Muhammad},
  title = {Autonomous Mango Harvesting Using YOLOv8 and ROS 2: A Simulation-Based Study},
  journal = {Food and Agriculture Communications},
  year = {2026},
  volume = {3},
  number = {1},
  pages = {23--33},
  doi = {10.62887/fac.003.01.00113}
}
Kostenlos starten →