Coordinated Control for Robot-assisted Take-off and Landing of Flying Robots

Kurzfassung

Up to date, state-of-the-art controllers for flying robots or Unmanned Aerial Vehicles (UAVs) are not prepared for tasks involving physical interaction, dynamically varying payload, or coordination with other robotic systems. For example, take-off or landing on a moving platform, e.g. a ship or a mobile robot, is challenging. The flying robot has to synchronize its movements to the motion of the platform. An estimate of the latter is uncertain or may not be available at all. The UAV has to react quickly, which is energy consuming and can lead to actuator saturation. This gets even worse if the aerial vehicle carries a payload or is under the influence of wind and turbulences. To ensure safe and robust take-off and landing, assistance systems for UAVs have been proposed in the literature. However, the available approaches provide insufficient support for the flying robot, lack a docking interface, do not take the distributed control system into account, and are only concept studies. In this work, a robotic assistance system for take-off and landing of flying robots is developed. The support system is based on a robot manipulator mounted on the landing surface. It allows to autonomously release or capture a UAV and to perform assisted take-off or landing. The main contributions of this thesis are the development of suitable dynamics models and coordinated control approaches for flying robots and robot manipulators, the realization of an assistance system prototype, and the implementation and evaluation of the presented controllers in simulations and experiments. Regarding the realization of the support system, a universal hinge mechanism for the connection between flying robot and robot manipulator is introduced. It leaves the rotational degrees of freedom of the flying robot open and results in a redundant system. It is shown that the redundancy allows to fulfill additional objectives, for example to reduce the workload of the manipulator. In this case, both systems jointly contribute to the assistance task, which is especially useful for heavy UAVs. Capturing the flying robot by means of the robot manipulator requires accurate target tracking. Suitable localization systems and a sensor fusion algorithm are evaluated and different docking interface designs are discussed. The developed coordinated controllers consider the different dynamical properties and actuation principles of flying robots and robot manipulators. In particular, three dynamics representations are used yielding controllers with varying complexity and performance: the independent dynamics, the combined dynamics, and the decomposed dynamics of flying robot and manipulator. To provide safe physical interaction, the well-known concept of impedance control is utilized. For flying robot control, established control laws are combined with external wrench observers, a novel control allocation procedure, and a novel adaptive control approach. This results in increased robustness against contact forces, actuator saturation, and changes in atmosphere or payload while providing more accurate trajectory tracking, as verified in the experiments. For a quadrocopter with fully symmetric fixed-pitch propellers, this work shows the first autonomous flight using bidirectional thrust which drastically increases control authority. This demonstrates the potential of optimized hardware in a combined assistance system.