Traditional safety standardization, as given by IEC 61508 and ISO 26262, is based on the V-model paradigm. It ensures, that requirements are fulfilled and the resulting system is ready for its dedicated use. This is particularly important in the certification process. Autonomous systems, such as mobile robots and self-driving cars, rely on AI and are considered safety critical systems, as they are used in safety relevant contexts. The key question is how to develop an AI-based safety-critical system. The development process of AI-based models shows an iterative structure. This process is repeated until a trained model passes all prescribed tests. However, this approach is inverse to the standard V-model and thus, incompatible for safety engineering. In this context, standardization can assist in making emerging technologies, such as AI, possible, by defining a development framework, demonstrating AI and functional safety compliance. In this document we investigate ISO/IEC TR 5469, the first step towards standardizing AI for FuSa on an international level. We discover that the central question is still unresolved. Hence, this manuscript motivates further endeavors to provide solutions that make AI-based safety-critical systems possible.
Ensuring functional safety and cybersecurity in automotive applications is a pressing concern in nowadays. The complexity of systems with safety-critical functionalities poses a challenge for effective verification approaches to reduce the gaps between safety and cybersecurity. Semiconductors are pivotal to automated driving systems due to their role in implementing complex functionalities. Therefore, it is essential to consider both Functional Safety (FuSa) and Cybersecurity aspects when developing semiconductor devices. This paper presents a holistic approach to incorporating failure effects and modes that considers both FuSa and Cybersecurity. Additionally, it investigates the potential relationship between failures effects and/or attack effects in semiconductors and emphasizes the interplay between FuSa and cybersecurity in an analysis. By illustrating proposed co-analysis concept between these areas, this study aims to provide insights into ensuring comprehensive FuSa and Cybersecurity measures in semiconductor development for automotive applications.
Operational safety assurance for autonomous and highly automated systems is imperative due to the inherent complexities and dynamic operational environments these systems encounter. As autonomous systems evolve, ensuring their safety throughout their life-cycle becomes indispensable, not just during the deployment but also during operation. Safety assurance has evolved from traditional document-based approaches to more structured, model-driven methodologies to enhance clarity and traceability. This evolution has been driven by the need for more rigorous, repeatable, and scalable processes to ensure safety in increasingly complex systems. Traditional safety assurance methods do not adequately address the dynamic real-time challenges and changes an autonomous system might face, necessitating a more adaptive and continuous approach. This paper highlights the need for a Cumulative Operational Safety Assurance approach addressing multiple research gaps. The cumulative aspect of the approach ensures that the safety assurance case evolves continuously and incrementally with operational data, improving decision-making and system safety.
The use of unmanned aerial vehicles (UAVs) is rapidly increasing. Certain tasks necessitate the deployment of multiple UAVs for a unified mission, forming what is known as a UAV swarm. A critical factor in assessing a UAV or UAV swarm is reliability. Various mathematical models are employed to analyze reliability. For instance, Binary-State Systems (BSS) models evaluate two states: operational/functional and faulty. Some research advocates for the use of a Multi-State System (MSS) model. MSS enables the examination of more than just two states (operational and faulty), providing a more comprehensive analysis. This paper presents a comparative study on calculating the availability of a UAV swarm using both BSS and MSS models. Different UAV swarm topologies are interpreted as typical MSS structures (series, parallel, k-out-of-n) and represented by structure function. The analysis includes both homogeneous and heterogeneous UAV swarms, which can be either non-redundant or redundant hot stable systems.
The rapid development of the low-altitude economy has propelled research focus towards technology pertaining to Unmanned Aerial Vehicle (UAV) swarms. The operational environment of UAV swarms is fraught with uncertainties and risks, posing challenges to their reliable functioning and mission success rate. UAV swarms possess the resilient property due to its self-organization and self-adaptation capabilities. Resilience design and enhancement serve as an effective approach for complex network systems like UAV swarms to effectively tackle unknown risks. The foundation of resilience design and improvement lies in the utilization of scientific and rational resilience analysis methods. The current three types of system resilience analysis methods effectively address the issue of describing resilience phenomena, but they have not delved into the level of understanding the laws governing resilience in UAV swarms. In this paper, the concept of intrinsic resilience (IR) is proposed for the first time to demonstrate the inherent resilient performance of the UAV swarm itself. The IR parameters including flexibility, rebounding velocity and restoration rate are provided and a 7-tuple metric framework is established. A UAV delivery system is utilized as an example to explain and illustrate the proposed IR analysis method.
