Terrain Avoidance Model Predictive Control for Autonomous Rotorcraft

This paper presents a terrain avoidance control methodology for autonomous rotorcraft applied to low altitude flight. A model predictive control formulation is used to adequately address the terrain avoidance problem, which involves stabilizing a nonlinear highly coupled dynamic model, while avoiding collisions with the terrain and preventing input and state saturations. Computing the model predictive control law amounts to solving a finite horizon open-loop optimal control problem subject to the state difference equations that describe the rotorcraft nonlinear dynamic model. State and input saturations are added to the optimization cost functional as penalties and terrain avoidance is achieved by penalizing the distance between the vehicle and the closest point on the terrain, yielding smooth and collision-free trajectories. Simulation results, obtained with a simplified version of a small-scale helicopter nonlinear dynamic model, are presented to assess the performance of the methodology with different reference paths and terrain profiles, including the extreme case where a desired path leads to collision with the terrain.