Sistema d'ajuda multisensor a la navegació per drons multiplataforma

Currently, and since the beginning of the millennium, drone technology is in a continuous growth, from which different techniques have been developed making it more accessible and easy to use in a variety of work areas, such as archaeology. In this field, a new project proposal is being developed by Hemav Foundation, in which is intended to used drones and image processing techniques to substitute the traditional pottery detection method and the subsequent analysis. In order to do it, the vehicle must carry on a multispectral camera and fly the study area in an autonomous way with the help of a GPS, which means, with no pilot, and so it needs some navigation helps. Flight altitude of the mission should be low enough to get high-resolution images and should keep it constant to not influence the area-pixel ratio. With this in mind, two navigation help systems are needed: one in order to keep altitude constant and follow strictly the terrain shape it is flying and another one in order to detect and avoid obstacles due to low altitudes imply more collision probability. The aim of this work is the design and the implementation of an altitude maintenance system and an obstacle detection and avoidance system. First a market study is done, in which both the methodologies that can be used and the devices that are available currently in the market are analysed, in order to finally choose which are the more suitable ones for the two systems in design. Given the actual situation, the sensor that was selected to measure altitude, the LIDAR-Lite v3, wasn¿t able to use and so another one has been used: TF mini, which specifications are clearly worse than the other ones. For obstacle detection camera D435 has been used. After that, the systems design is done. On one hand, the hardware architecture: connection between camera and flight controller, Pixhawk Cube. On the other hand, the software: choose a programming language, which will be Python, download and install the required libraries for obtaining data (Intel RealSense SDK) and for communicating with the drone and between internal components (MAVLink and DroneKit) and, finally, to code the program. After all, some test has to be done: first the sensor and the camera to check the specifications and the code to check the behaviour. On one hand, the sensor has good accuracy but it takes some seconds to stabilize and the maximum range is not enough. On the other hand, the camera works so well: it detects obstacles of different shapes and materials and situated in different locations. Finally, the code works, although it could be improved in terms of efficiency. Summarizing, both systems design can be implemented and used in the project if the altitude sensor is changed by another one with higher maximum range, lower accuracy and a more stable behaviour