Answers are grounded in this paper and cite their sources.
Orchard monitoring increasingly combines an aerial survey by an unmanned aerial vehicle (UAV) with closer inspection of selected trees by unmanned ground vehicles (UGVs). Each inspection must then be assigned to a robot that carries the right sensor and has a feasible route and enough energy. In this work, mission assignment is expressed through Kubernetes. Robots are represented as Nodes and missions as Pods with versioned requirements, while mission feasibility is checked, eligible robots are ranked by battery state and target proximity, and mission capacity is reserved by a Scheduling Framework plugin. The plugin was evaluated over 30 paired runs in a simulated 800 m by 800 m orchard served by one UAV and seven UGVs. Comparisons were made with the Kubernetes 1.34.1 default scoring profile and three reference policies under identical feasibility checks. Relative to the default scoring profile, the mean route per completed mission was shortened by about 91 m with the agricultural mission profile, and the arrival of the first robot at the 20% battery threshold was delayed by about 29 min. Consistent results were obtained from a reanalysis at a common time horizon and from a study of the battery and proximity weights. Slightly shorter routes were obtained with route oriented policies and the latest threshold with balanced weighting. Binding and mission dispatch were confirmed by integration tests on a local multi-node Kubernetes cluster, in which a mock ROS 2 action runtime stood in for the robots. Field calibration of energy, routes, and sensing is the next step.
No takes yet. Share an insight, caveat, or question.
George Koufoudakis (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: