Autonomous systems are on the rise. However, the challenge to test autonomous systems to ensure their safe and fault‐free behaviour is not solved yet. This is especially critical when we consider the fact that autonomous systems are often safety‐critical systems envisaged to interact with humans without explicit human supervision. This paper points out why testing autonomous systems is such a challenge and provides an overview of the current state‐of‐the‐art theory and practice. The gathered information is then condensed into guiding points for the way forward.

Modern avionics systems are characterized by their increasing complexity and demanding safety requirements. The fulfillment of safety constraints is often only tested after the design of the systems, which leads to costly and time‐consuming design iterations. The complexity of avionics systems is still managed through a number of predominantly manual, paper‐based procedures. Relying on systems engineering and operations research methods and tools, a methodology was developed to support the design process. This methodology should reduce the design duration through an earlier consideration of safety constraints. It should also contribute to the overall improvement of the design, through an automated approach of its complexity. A generic meta‐model is created, suitable for various kinds of avionics systems, which contains requirements ‐ including safety requirements ‐ and design goals. From these formal requirements and goals an algorithm creates corresponding mathematical constraints and optimization criteria. All valid architectures fulfilling these constraints are automatically generated. Sets of efficient and diverse architectures among the generated solutions are selected by an algorithm.

No abstract

No abstract

No abstract