Abstract

Unmanned vehicles have been increasingly employed in real life. They include unmanned air vehicles (UAVs), unmanned ground vehicles (UGVs), unmanned spacecrafts, and unmanned underwater vehicles (UUVs). Unmanned vehicles like any other autonomous systems need controllers to stabilize and control them. On the other hand unmanned systems might subject to different faults. Detecting a fault, finding the location and severity of it, are crucial for unmanned vehicles. Having enough information about a fault, it is needed to redesign controller based on post fault characteristics of the system. The obtained controlled system in this case can tolerate the fault and may have a better performance. The main focus of this thesis is to develop Fault Detection and Diagnosis (FDD) algorithms, and Fault Tolerant Controllers (FTC) to increase performance, safety and reliability of various missions using unmanned systems.

In the field of unmanned ground vehicles, a new kinematical control method has been proposed for the trajectory tracking of nonholonomic Wheeled Mobile Robots (MWRs). It has been experimentally tested on an UGV, called Qbot. A stable leader-follower formation controller for time-varying formation configuration of multiple nonholonomic wheeled mobile robots has also been presented and is examined through computer simulation.

In the field of unmanned aerial vehicles, Two-Stage Kalman Filter (TSKF), Adaptive Two-Stage Kalman Filter (ATSKF), and Interacting Multiple Model (IMM) filter were proposed for FDD of the quadrotor helicopter testbed in the presence of actuator faults. As for space missions, an FDD algorithm for the attitude control system of the Japan Canada Joint Collaboration Satellite - Formation Flying (JC2Sat-FF) mission has been developed. The FDD scheme was achieved using an IMM-based FDD algorithm. The efficiency of the FDD algorithm has been shown through simulation results in a nonlinear simulator of the JC2Sat-FF.

A fault tolerant fuzzy gain-scheduled PID controller has also been designed for a quadrotor unmanned helicopter in the presence of actuator faults. The developed FDD algorithms and fuzzy controller were evaluated through experimental application to a quadrotor helicopter testbed called Qball-X4.