Extension of a Deterministic Path Planner for Cooperative Multi-UAV Missions through the Use of Modern Artificial Intelligence Methods

This work investigates a deterministic path planner for Unmanned Aerial Vehicles (UAVs). This planner controls UAVs and is intended to survey a search area using Angle of Arrival (AoA) measurements. Both the environment and the mission requirements can change dynamically. To increase adaptability, a preceding neural network is integrated. The neural network is trained previously and dynamically adjusts six configurable parameters (𝛼1, 𝛼2, . . . , 𝛼6) of the path planner. In this way, the planned UAV paths can be influenced. However, the behavior of the path planner remains deterministic. This means that certification in aviation according to established standards and procedures would still be possible.