Use of Unmanned Aerial Systems in Civil Applications

Interest in drones has been exponentially growing in the last ten years and these machines are often presented as the optimal solution in a huge number of civil applications (monitoring, agriculture, emergency management etc). However the promises still do not match the data coming from the consumer market, suggesting that the only big field in which the use of small unmanned aerial vehicles is actually profitable is the video-makers’ one. This may be explained partly with the strong limits imposed by existing (and often "obsolete") national regulations, but also - and pheraps mainly - with the lack of real autonomy. The vast majority of vehicles on the market nowadays are infact autonomous only in the sense that they are able to follow a pre-determined list of latitude-longitude-altitude coordinates. The aim of this thesis is to demonstrate that complete autonomy for UAVs can be achieved only with a performing control, reliable and flexible planning platforms and strong perception capabilities; these topics are introduced and discussed by presenting the results of the main research activities performed by the candidate in the last three years which have resulted in 1) the design, integration and control of a test bed for validating and benchmarking visual-based algorithm for space applications; 2) the implementation of a cloud-based platform for multi-agent mission planning; 3) the on-board use of a multi-sensor fusion framework based on an Extended Kalman Filter architecture.