Edge-AI-Based Vision-Guided Robotic System for Selective Trimming of Unhealthy Leaf Clusters in Controlled Agricultural Environments
Document Type
Conference Proceeding
Publication Date
1-1-2026
Abstract
This paper presents a low-cost edge-AI-based robotic platform for selective detection and trimming of unhealthy leaf clusters in controlled agricultural environments. The system combines a four-wheel-drive mobile base, a 3D-printed 4-degree-of-freedom robotic arm, a Raspberry Pi 4 running a YOLOv8-based vision model, and an Arduino Uno for low-level control. The robot performs an onboard perception-to-action cycle in which unhealthy clusters are detected, localized from bounding-box centroid information, mapped to manipulator coordinates, and removed using inverse-kinematics-based arm motion. Experimental evaluation over 50 autonomous test cycles showed 91% detection accuracy, 87% trimming success, 92% path-tracking accuracy, and 96% base-detection reliability. The results demonstrate the feasibility of affordable mechatronic hardware and embedded AI for selective plant-level intervention in sustainable agriculture.
Publication Title
2026 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies Aeect 2026 Proceeding
ISBN
[9798319529602]
Recommended Citation
Ryalat, M.,
Almtireen, N.,
Al-Refai, G.,
Elmoaqet, H.,
&
Rawashdeh, N.
(2026).
Edge-AI-Based Vision-Guided Robotic System for Selective Trimming of Unhealthy Leaf Clusters in Controlled Agricultural Environments.
2026 IEEE Jordan Conference on Applied Electrical Engineering and Computing Technologies Aeect 2026 Proceeding, 278-284.
http://doi.org/10.1109/AEECT69724.2026.11657831
Retrieved from: https://digitalcommons.mtu.edu/michigantech-p2/3009