In agricultural field mapping missions, a collision risk occurs when the UAVs deviate from their planned trajectories due to the wind or uncertainties in the model, but also in case of intersecting paths of the UAVs, e.g. when the battery level is not sufficient and a UAV needs to change its initially planned path and return to the base unexpectedly.In this paper, three nonlinear model predictive control (NMPC) trajectory tracking strategies for collision avoidance are compared for multi-UAV mapping of an agricultural field: incorporating collision avoidance as a nonlinear constraint, applying it as a penalty cost, and employing a safe flight corridor approach. All the presented strategies consider passing priority allocation of the UAVs involved in the mission, where only a UAV with a lower-level priority handles collision avoidance. Control strategies are compared regarding robustness to external disturbances, such as wind, and model uncertainties through Monte Carlo simulations. The performance is evaluated with respect to the resulting tracking errors, the ability to avoid collision, and the computational time needed to solve the optimal control problem. The objective is to determine, among these three approaches, the one that exhibits the best trade-off between performance and computational burden.

This paper presents a novel approach for UAV battery management of the energy-aware mapping mission planning. The optimization strategy aims to reduce overall flight time by considering necessary battery replacements. Energy constraints are imposed as the capacities of available batteries are limited. The resulting mission plan reduces the unnecessary long flight distances from and to the base station. This strategy also considers the choice of the base location. Potential base locations are fixed and placed at the vertices of the mapping area for easy access. Iterative optimization results in choosing the location with the minimal total mission flight distance. The optimal solution of the proposed approach is compared to its initial solution, where the mapping area is decomposed proportionally to the capacities of batteries planned to be used in the mission. Simulation results for the realistic test case mapping area of the vineyard and olive orchard show that fewer battery replacements are needed and both total flight time and total flight distance are reduced when implementing the proposed optimization strategy. Moreover, mission safety is increased as the number of take-offs and landings, which pose high risk, are reduced.