Trajectory Generation for UAVs in Unknown Environments with Extreme Wind Disturbances (1906.09508v1)

Abstract: The widespread use of unmanned aerial vehicles (UAVs) by the military, commercial companies, and academia continues to push research for autonomous vehicle navigation, particularly in varying environmental conditions and beyond-line-of-sight (BLOS) applications. This article addresses trajectory generation for UAVs operating in extreme environments where the wind disturbances may exceed the vehicle's closed-loop stability bounds. To do this, a controller is developed that has two modes of operation: (1) normal mode, and (2) drift mode. In the normal mode the vehicle's thrust and sensor limitations are not exceeded by environmental conditions, whereas in the drift mode they are. In the drift mode, a drift frame that moves with the prevailing wind is established in which the vehicle maintains control authority to generate and track trajectories. The vehicle maintains control authority by relaxing the inertial frame trajectory tracking requirement and re-planning the trajectory in the drift frame. Guarantees are established to ensure tracking of the trajectory, collision avoidance, and respecting the vehicle thrust and sensor limitations. Simulation results demonstrate the algorithm properties through two scenarios. First, the performance of two quadrotors is compared where one utilizes the drift mode and the other does not. Second, multiple vehicles navigate through two narrow openings between protected and windy environments to demonstrate on-board updates to navigation parameters based on environmental conditions.