Three-Dimensional Open-Path Sequencing for Robotic Tea-Shoot Plucking via Hierarchical Ant Colony Optimization and Endpoint Coordination
Year: 2026
Type: article
Abstract: Efficient three-dimensional (3D) open-path sequencing is important for robotic selective plucking because harvestable tea shoots are irregularly distributed in canopy space. This study formulates tea-shoot plucking as an open-path sequencing problem over preparatory plucking points and proposes a hierarchical ant colony optimization framework with endpoint-aware traversal-direction coordination. Preparatory points are decomposed into local operating regions, within which routes are optimized using ant colony optimization and 3-opt refinement. Centroid-level ant colony optimization then determines the cluster visiting order, followed by dynamic programming to select the traversal direction of each local route according to endpoint-to-endpoint connection costs. Across eleven simulated scenarios, including an unstructured Global-random control, the proposed method achieved the shortest mean route length in ten scenarios. In the Large-scale scenario, it achieved a mean route length of 13.973 ± 0.184 m, which was 10.5% and 2.9% lower than those obtained by global ant colony optimization and greedy insertion, respectively. On five real tea-field point-cloud datasets, the proposed method reduced mean route length by 0.57–3.15% relative to greedy insertion, with statistically significant improvements in four datasets. A preliminary CoppeliaSim assessment further showed that the optimized visiting sequence could be geometrically connected using smoothed local transfer trajectories. Overall, the proposed framework improves sequence-level route quality and computational efficiency for clustered 3D tea-shoot distributions, while its field-level operational benefits remain to be established through closed-loop robotic validation.
Source: Agriculture
Institutions: Jiangsu University, Yunnan Agricultural University
Language: English
Topic: Smart Agriculture and AI
Subfield: Plant Science
Domain: Life Sciences
Open Access status: gold
Estimated APC paid: $3,137
Award: G66720820
OpenAlex ID: W7207862012