Small unmanned helicopter autorotation using non-linear model predictive control

Small unmanned helicopters are suitable for a variety of applications including search and rescue, surveillance, communications, traffic monitoring as well as inspection of buildings, power lines and bridges. This paper presents an on-line, model-based, real-time autonomous autorotation controller, tailored for small helicopters. The approach selected is based on non-linear model-predictive control techniques that allow constraint handling and provide relatively good performance. The non-linear optimization is handled by a specially-designed recurrent neural network so that fast convergence is achieved. The goal of this controller is to eliminate the risk to people on the ground for failures that can be accommodated via autorotation. This is accomplished by appropriately lowering the kinetic energy of the helicopter during the last phase of its descent.