Autonomous Driving Systems (ADS) are considered safety-critical, as even a minor fault may lead to catastrophic consequences. To evaluate their reliability and robustness under failure conditions, Fault Injection (FI) techniques have been widely adopted. Most existing FI methods employ data-driven approaches, such as surrogate modeling and reinforcement learning, to generate test cases. While these techniques have shown promise, they often incur substantial costs in terms of data collection and training time. Moreover, their performance is highly sensitive to the quality and quantity of training data, which can limit their applicability in diverse or unseen scenarios. In this paper, we propose LOFT, an efficient multi-objective search-based FI testing framework that leverages Large Language Models (LLMs) to identify diverse and realistic critical faults. To accommodate the structured and non-linguistic nature of raw simulation data, LOFT adopts a two-stage LLM-based fault injection pipeline. In the first stage, an LLM converts singleframe simulation data into natural language descriptions and suggests appropriate fault types. In the second stage, a separate LLM examines the broader scenario context to determine the optimal time window for fault injection. The outputs from the two LLMs are then used to initialize and guide a multi-objective search procedure aiming at discovering a diverse set of critical faults. We implement LOFT and evaluate on an ADS provided by our industrial partner. Experimental results show that, compared with two baseline approaches, LOFT detects over $90 \%$ more critical faults and identifies an average of 2.2 additional fault types within an equivalent number of simulations.

The progressively dominant role of software in safety-critical systems raise concerns about the software dependability. There are limited mature practices and guides for assessing software dependability and analyzing system-level hazards triggered by software anomalies. A problem is that faults, errors, and failures that represent software anomalies, albeit with different natures, are usually used indistinctly to predict software dependability, leading to unsolid results. The lack of such consensual conceptualization also leads to poor interoperability between supporting tools, and, consequently, difficulties in anomaly management and software maintenance. Anomaly analysis and management is more tough for safety-critical software due to its higher complexity and the safety-critical nature. The complex context of safety-critical software causes difficulties in determining the evolution/propagation path of software anomalies and the impact on system safety. To capture the nature of safety-critical software and support an understanding of mechanisms of software anomalies and associated hazards, we propose three reference ontologies: Safety-critical Software Ontology, Software Fault Ontology and Software-failure-induced Hazard Ontology, which are built based on international standards, guides, and relevant conceptual models. We also discuss the relationships among them. That will facilitate a better understanding of the software anomaly mechanisms and the design of intervening/mitigation solutions. We demonstrate how these ontologies can help analyze software problems of real-world safety-critical systems.