Supermaneuverable perching

Abstract

Birds have the impressive ability to gracefully 'swim' through the air while executing aerobatic maneuvers that routinely defy modern aeronautical and control engineering, consistently reminding us that the skies are truly their playground. These animals are masters at inducing and exploiting post-stall aerodynamics to quickly execute maneuvers with unprecedented precision, with nowhere near the sustained propulsive power found in modern state-of-the-art aircraft. This amazing ability to manipulate the air is commonly attributed to the intricate morphology of the wings, tail, feathers and overall sensory motor system of the animal. In this thesis we demonstrate, on real hardware, that using only an approximate model of the post-stall aerodynamics in combination with principled and novel tools in optimal control, even a simple fixed-wing foam glider (no propeller) made out of rigid flat plates, with a single actuator at the tail, is capable of executing a highly dynamic bird-like perching maneuver to land on a power-line by exploiting pressure drag on its stalled wings and tail. We present a feedback controller capable of stabilizing the maneuver over a wide range of flight speeds and quantify its robustness to wind-gust disturbances. In order to better characterize the aerodynamics during perching, we performed smoke-visualization in a low-speed free flight wind-tunnel, where we were able to capture real images of the dominant vortex wake dynamics. We describe the application of these results to the synthesis of higher fidelity aerodynamic models. We also demonstrate our initial perching experiments with flapping wings, using a flapping-wing version of our glider as well as our fully computerized two-meter wingspan robotic bird, Phoenix.