Immune-inspired fault diagnosis for a robotic system

Abstract

To achieve fully autonomous systems, fault tolerance is often employed. Fault tolerance is the ability to continue operation in the presence of faults. Fault diagnosis is an essential component of fault tolerance, especially for autonomous robotics. It is the process of determining as much information as possible about the fault, especially the origin of the fault. However, a real time fault diagnosis for resource limited robotic systems has proposed a new set of challenges, such as its complexity and efficiency, which traditional methods will find difficult to meet. This has led the work to seek inspiration from the immune system, where an effective and efficient fault diagnosis solution has been provided for thousands of years. This thesis presents a novel immune-inspired on-line fault diagnosis algorithm for robotic systems and includes the first application of that Artificial Immune System to robot fault diagnosis.