| dc.description.abstract | In recent years, the commercialization and militarization of unmanned arial vehicles (UAVs) have become increaseingly common, with quadcopters being particularly prevalent in various applications. Despite being inherently unstable, coupled, and underactuated systems, quadcopters benefit from their low cost and relatively easy controllability, which often makes them a tool for validating control theories or implementing applications.
Due to the aforementioned characteristics of the system, if a propeller blade is damaged or a motor fails, the aircraft may lose certain degrees of freedom and become uncontrollable in certain attitudes. The main approaches to address this issue are collision avoidance and control under failure conditions. The first approach typically involves equipping the aircraft with distance sensors at appropriate locations on the fuselage, combined with algorithms for obstacle avoidance. The second approach requires identifying the failure model first and then implementing control based on the current model.
This study investigates the partial failure of a single motor in a quadcopter. Motor failure is simulated during flight, and failure identification is conducted using only attitude data and an intuitive method of determining the sign of the rotational error. P-PID control and the coupling characteristics of the quadcopter are employed for fault compensation and control. The goal is to achieve fault control while maintaining the aircraft at a certain altitude and without the yaw degree of freedom. At an altitude of 4 meters, with remaining motor efficiencies of 0.8, 0.6, and 0.2, the tilt angle can be controlled within 0.2 radians before landing, and the impulse upon ground contact is kept within 3.2 (kg ‧ m/s). This ensures that the fuselage lands slowly in a horizontal attitude, allowing the impact to be distributed across the four arms.
In the experimental phase, the system is initially simplified through assumptions to simulate the identification and control process from normal flight to failure. Subsequently, the control methods are verified through actual outdoor flights. | en_US |