Although deep neural networks have proven to be successful across a large variety of machine learning tasks, recent work has demonstrated that they are at the same time vulnerable to so-called adversarial examples: inputs that are almost indistinguishable from natural data but misclassified by the network. In the case of image classifiers, such adversarial examples have traditionally been constructed by perturbing the original images, but ...

The goal of this bachelor thesis is to enable a special unmanned aerial vehicle (UAV) to make a transition from flying in the air to driving on a wall. For this, the Voliro hexacopter has been used as platform. Thanks to its tiltable rotors it can produce forces and moments in all directions and therefore hover in any arbitrary orientation. It is able to drive on the wall by flying vertically while at the same time pressing itself against ...

The ”Voliro” system developed during a one-year student project at ETH Zurich, is a novel concept of a rotary-wing UAV, designed as a multicopter with a classical planar configuration, whereby the thrust motors can be tilted around their axes. This allows the system to decouple position and orientation as well as move omnidirectionally. This bachelor thesis adapts the allocation method used within the project to tackle specific critical ...