In the past decade, Unmanned Aerial Vehicles (UAV) and Micro Air Vehicles (MAV) have increasingly become a topic of interest in different research organisations and industries. UAVs are being applied in various areas such as crop field observation, inspection of transportation infrastructure and package delivery. Currently, commercially available platforms are either expensive high-quality platforms or inexpensive products that do not offer the sufficient sensor quality required for research. A low-cost and modular multirotor solution is developed in this thesis that supports semi-autonomous flight. We employ first-person view (FPV) racing quadcopter base together with Pixracer, a low-cost and high-performance open hardware autopilot. The autopilot incorporates advantages of popular open source firmware, such as APM and PX4.
Additionally, a powerful computer with a low-weight and capable vision sensor is included in this configuration. It is shown, that the designed multicopter can effectively operate in manual mode, as well as perform autonomous flights by the use of either a motion capture system or a global navigation satellite system (GNSS). The collected visual data is used to create a 3D reconstruction of the environment using COLMAP, an open-sourced reconstruction tool. Reconstructed scenes can be used for example in later visual inspection or environment measurement tasks.
|
