|Title:||A Taxonomy of Safety Related Challenges in Using Autonomous Systems in Safety-Critical Applications|
Human factor has been regarded as one of the main causes of accidents in safety-critical systems. To reduce the risk of accidents due to human factors, the modern safety-critical systems are being developed with more and more control, making them increasingly autonomous. Unlike for traditional safety-critical systems, the system behavior of these autonomous systems cannot be fully defined during design-time. Such systems are often designed to learn and adapt during runtime using Artificial Intelligence (AI). Safety-critical functions based on AI are difficult to assure using the traditional safety assurance principles that do not take into account the dynamic nature of these systems. For example, the challenge of assuring an autonomous car is similar to the challenge of assuring that it is acceptably safe for a human driver to operate a car. The approval cannot be simply issued once, but has to be constantly evaluated. Just as we constantly monitor human drivers to ensure that it is adequately safe for them to drive a car, so we need to constantly monitor and assure that an autonomous car is still acceptably safe to operate.
The focus of research on autonomous safety-critical systems has been increasing lately and many industries are struggling with adopting such autonomous systems in safety-critical applications. A taxonomy of safety challenges related to autonomous safety-critical systems and the possible ways to address these challenges is required to ease the adoption of autonomous systems in safety-critical applications. To build such taxonomy, we propose to make a survey that should be cross-domain, e.g., including such application domain as health, automotive. The aim is to analyze results further statistically and provide mapping e.g., to application areas, related standards, publication dates.
The following research questions (RQs) are formulated to address the specified goal:
RQ1: Which autonomy levels have been in focus of research in the safety domain in the past years?
RQ2: What are the safety-related challenges identified for using autonomous systems in safety critical applications?
RQ3: What are the safety measures proposed to address the identified challenges?
We propose to do a systematic literature review (SLR) that aims to provide answers to the formulated above RQs. The results of the SLR will be analyzed to provide a taxonomy of presented challenges related to safety of autonomous systems. A SLR is a well-defined process of extracting information, thus we propose to follow guidelines regarding the process from B. Kitchenham and S. Charters .
 B. Kitchenham and S. Charters, “Guidelines for performing systematic literature reviews in software engineering version 2.3,” EBSE Technical Report, Keele University and University of Durham, 2007.
|IDT supervisors:||Elena Lisova, Irfan Sljivo|