Quadcopter Dynamics Analysis in Real Life and Simulation

Aquest ítem conté el treball original, defensat el 27 de gener de 2020, així com una versió corregida del mateix a data de 26 de febrer de 2020. Els canvis introduïts a la segona versió consisteixen en correccions estilístiques i de tipografia principalment.Helicopters are flying machines equipped with a powerful rotor which pushing air downwards provides the thrust needed to fly. One rotor is not enough to guarantee the control of the helicopter because it would be impossible to counter the torque it produces thus leading the fuselage to spin in the opposite direction. Many solutions have been carried out in the last century but all requiring a complicated design of the rotors and other mechanisms. A quadrotor helicopter (quadcopter) is a helicopter which has four coplanar rotors typically mounted at the vertices of a X-shape body that provide thrust. Using this new configuration, it is possible to balance the torques produced by the rotors and control the flight acting just on their rotational speeds, avoiding moving parts. The main challenge in projecting quadcopters has been controlling six degrees of freedom acting on four independent rotors. In the past few years, due to the decreasing cost of new microprocessors able to compute highly nonlinear dynamics equations, quadcopters are becoming increasingly popular. In this thesis I developed a comparison between the performances of a real quadcopter (the DJI Tello Edu) and its pc simulation (using the software Airsim). More specifically I got the drone parameters through a series of tests, I built a custom version of the simulation software using the previously obtained data, I developed the same flight plan for both real and simulation drone and at the end, I made a comparison between them. The results of this study will be useful in order to develop a Deep Reinforcement Learning method on the simulation drone and then transfer it on the real on