Model-Based Design of a Control System for a Group of Agricultural Drones and a Robotic Swap Station

Abstract

This article investigates the principles of designing a control system for a group of agricultural spraying drones and a robotic service station for automatic refueling and battery swap. The system concept is based on a hybrid approach combining elements of centralized and decentralized control: global planning of routes and flight schedules is performed centrally, while local trajectory corrections during flight are permitted at the individual drone level. The response to abnormal (emergency) situations is distributed between central and local control levels. The technical implementation of this concept is expediently realized using a multi-agent architecture that includes drone agents, a service station agent, and an external control agent. The study develops a mathematical framework to quantify drone route deviations, considering both spatial cross-track errors and temporal scheduling offsets. Motion monitoring criteria and admissible deviation limits are introduced, with their exceedance triggering emergency operating modes. Parameters for damping temporal mismatches during approach to the station are defined: for drones arriving ahead of schedule, holding points with battery energy constraints are provided; for delayed drones, a reserved waiting interval between service operations is utilized. A simulation methodology is proposed for validating guidance algorithms and failure criteria at the conceptual design stage. Overall, it is shown that model-oriented design reduces the risk of errors during physical implementation of a complex robotic system and decreases development costs.