Cloud-based mission coordinator for Unmanned Aerial Systems (UAS) in urban applications

In the last years, the use of Unmanned Aerial Systems (UASs) has experienced a significant increase. They are becoming more affordable and accessible and many industries are attracted by the benefits that Unmanned Aerial Vehicles (UAVs) can offer. One of the most promising UAV applications rests on the logistic industry, specifically the delivery of goods. Some companies are investing heavily for some time to create a delivery service based on these aerial vehicles. On the one hand, the scenario seems to be very appropriate. The predictions of urban population rise and the trend of cities development of being transformed into smart cities provide important ingredients for a successful implementation of UASs. On the other hand, this scenario presents difficulties originated by regulation constraints. So far, the execution of missions that involve unmanned aerial vehicles require a lot of staff. Against this fact, an autonomous cloud-based mission coordinator is proposed in this project. This coordinator will reduce the requirements of personnel and will help to perform predefined missions by a fleet of UAVs. A very important action that is carried out by the mission coordinator is path planning. The main work of this project is focused on this task which is crucial to obtain optimal performances. The path planner uses mission requirements, such as the starting point and the target point, and a model of the environment to compute and optimise paths that vehicles will follow to accomplish the mission. The environment has been modelled as a static cell-decomposed map in which each cell has an associated risk value. This value plays the role of a crossing cost while paths are computed. Research on optimisation techniques in this field has been done and three different methods (Dijkstra's algorithm, A* and D*) have been tested to obtain the path solutions. It was noticed that their performances depend on the features of the environment model, mainly on the size. In this approach, the management of a fleet of UAVs and the presence of docking stations are considered. These stations provide places where vehicles could be recharged or cargo(es) may be exchanged. The result is a multi-vehicle problem that must be solved by using the adequate resources of the system. For this reason, the mission coordinator also generates a tree with all the possible path combinations between current locations of vehicles, docking stations, and picking and target points to find the best group of vehicles to perform the mission. The optimal solution is represented by the branch of the tree that has the lowest accumulated cost. In order to simulate delivery missions and develop the mission coordinator a series of functions and programs were created with MATLAB. After being tested, a web application based on these MATLAB scripts was created and hosted in a local-server what provides a cloud-based architecture to the mission coordinator.