Motion estimation on UAVs

Depending on whether an Unmanned Aerial Vehicle (UAV) is in flight or on the ground, logical decisions are completely different. Firstly, if it can be assessed that a drone is still then its motors can be stopped, improving user safety and reducing unnecessary mechanical stress. It is also possible to calibrate inertial instruments in such conditions. This master’s thesis deals with two separate subjects: first, the implementation of an empiric algorithm in order to distinguish the flight status of an UAV, which means knowing whether the drone is in flight or on the ground. The consistency of the estimates at the root of the algorithm has been questioned and the second part of the thesis deals with the improvement of the state estimator of the UAV. Autonomous drones need accurate knowledge concerning their position, velocity and attitude estimates. The second part of the thesis describes the design and the implementation of a loosely coupled INS/GNSS system as a navigation system. Short term accuracy is provided by the Inertial Measurement Unit, whereas the long-term accuracy is guaranteed by Global Navigation Satellite System measurement updates. The characterization of stochastic errors was used in order to design the Extended Kalman Filter, and the proof of concept was established thanks to a simulation environment within MATLAB. The algorithm was implemented on an autopilot, the STM32F407. State augmentation can be achieved in order to improve the overall performance of the estimator.