A Real Time Expert Control System for Helicopter Autorotation

Abstract

Autorotation maneuvers are required to perform a safe landing of a helicopter in cases of engine loss in a single engine vehicle and transmission or tail rotor malfunction. The rise of autonomous helicopter technology, and the pilot skill required to manually perform an autorotation, motivate the need for new autonomous autorotation control laws. Previous approaches to automatic control for this maneuver have relied on control law optimization based on a high-fidelity model of the helicopter, or have attempted to match recorded trajectories flown by an expert human pilot. In this paper, a new expert control system is proposed. The term “expert control system” is used because the system is intended to mimic the actions that a human pilot might take, does not require any iterative learning, model prediction, or optimization at runtime, and is based on an inference system that involves fuzzy logic, PID, and other conventional control techniques. The multi-stage control law drives the helicopter to a near-optimal steady-state descent and uses an estimate of the time to impact to safely flare and land the helicopter in the vast majority of flight conditions. The control law is validated using a full 6-degree-of-freedom simulation of both a full-size attack helicopter and a small hobby-class helicopter. The pro- posed control design is highly flexible and may be used to perform fully autonomous autorotation or to provide guidance to pilots during manual autorotation maneuvers.