Trajectory Learning for Highly Aerobatic Unmanned Aerial Vehicle

Abstract

The dynamics of fixed wing planes are well understood within the conventional flight envelope. The situation is different in the case of aerobatic maneuvers with a large angle of attack, such as perching and vertical hover. In such maneuvers the airflows around the plane are unpredictable making it difficult to create accurate dynamic models, which would normally be needed for the design of conventional controllers. Yet human RC pilots are able to fly these maneuvers with fixed wing planes.

Apprenticeship learning provides a promising solution to the problem of automating highly aerobatic maneuvers. It allows the maneuver to be learned from demonstration flights done by a human RC pilot rather than relying on an accurate dynamics model.

The focus of this thesis is on a specific issue in apprenticeship learning, namely how to infer the trajectory the pilot intended to fly from a set of suboptimal demonstration trajectories. Such a trajectory can be used as a target trajectory for an autonomous controller. A trajectory learning algorithm that has shown promising results in automation of aerobatic helicopter flight is applied to a fixed wing UAV platform.

The algorithm is tested on two different maneuvers; A straight line of level flight, and a vertical hover maneuver. In the case of both maneuvers the algorithm learned the intended trajectory without prior knowledge about the trajectory.

In order to collect training data for the trajectory learning task, an appropriate platform and data acquisition system are needed. This thesis therefore also presents the development of a fixed wing UAV platform for research on automation of aerobatic flight.