Hierarchical fault detection and diagnosis for unmanned ground vehicles

This paper presents a fault detection and diagnosis (FDD) method for unmanned ground vehicles (UGVs) operating in multi agent systems. The hierarchical FDD method consisting of three layered software agents is proposed: Decentralized FDD (DFDD), centralized FDD (CFDD), and supervisory FDD (SFDD). Whereas the DFDD is based on modular characteristics of sensors, actuators, and controllers connected or embedded to a single DSP, the CFDD is to analyze the performance of vehicle control system and/or compare information between different DSPs or connected modules. The SFDD is designed to monitor the performance of UGV and compare it with local goal transmitted from a ground station via wireless communications. Then, all software agents for DFDD, CFDD, and SFDD interact with each other to detect a fault and diagnose its characteristics. Finally, the proposed method will be validated experimentally via hardware-in-the-loop simulations.