ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

Dual-Camera Vision Control and Manipulator Integration for Precision Apple Bud Thinning

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2026 ASABE Annual International Meeting  2601618.(doi:10.13031/aim.202601618)
Authors:   Kittiphum Pawikhum, Long He, Paul Heinemann
Keywords:   Apple bud detection, machine vision, deep learning, robotic thinning, crop load management

Abstract. Thinning is a critical practice in apple orchard management, directly influencing crop load and fruit quality. To assist automated crop load management, a global–local dual-camera vision control architecture was integrated with a robotic manipulator and a custom bud-removal end-effector for apple bud detection and thinning. Two YOLO-based object detection models were evaluated for early bud detection under natural lighting conditions. YOLOv12 achieved the best overall performance with a precision of 0.66, recall of 0.59, and mAP@50 of 0.61. Grad-CAM analysis further demonstrated YOLOv12‘s improved spatial focus on target regions. The dual-camera system enabled real-time 3D localization with RGB-D sensing to guide the manipulator for precise bud removal under field conditions. Field trials on 50 detected apple buds demonstrated effective system integration, with 44 successful removals, achieving an 88% removal success rate. Bud orientation emerged as the dominant determinant of removal outcomes, whereas kinematic simulation verified that the manipulator‘s workspace affords sufficient coverage and approach geometry for structured orchard rows. Overall, the study presents a scalable path toward precision bud thinning through coordinated global–local vision and robotic control.

(Download PDF)    (Export to EndNotes)