On September 27, Martin Schobesberger presented "A Survey on Safety II in Assistance Systems for Industrial Domains: State-of-the-Art, Advantages, Challenges, Recommendations" (Schobesberger M., Huber J., Haslgrübler M., Grünberger S., Hochreiter, D., Ferscha A., Effenberger G. und Malisa V.) at WOS (Working on Safety) 2022. Since 2002, the conference takes place every two years. This year's theme was "Focus on Humans in a Technological World".

Abstract

Digitalization and deployment of assistance systems in industrial domains are ever-growing – as is the change in safety paradigms from Safety I to Safety II. Therefore, the goal of this paper is a comprehensive study on such assistance systems containing (i) an analysis of the state-of-the-art with regard to implemented safety strategies, (ii) an investigation of advantages and challenges of Safety II based assistance systems and (iii) recommendations for the development process with focus on maximizing the advantages of Safety II while minimizing its challenges.
The paper is structured in the following way: In the first section a general introduction to safety paradigms and assistance systems with a special focus on industrial domains is given, followed by a survey on the use, realization, implementation, usability and experiences of Safety II in existing assistance systems. This survey also takes into account socio-technical aspects, the actor-network theory, intention recognition and AI in industrial environments. Based on this survey, the authors detected and extracted potential advantages as well as challenges in the implementation process of Safety II – both of which are be described in the section “Advantages and Challenges of Safety II in Assistance Systems”. As a result of this collection, the section “Recommendations” provides a comprehensive list of best practices, potential pitfalls and recommendations on the inclusion of Safety II aspects in industrial assistance systems. The authors also describe and discuss their own experiences on planning and proposing a framework for future Safety II based assistance systems. This proposed framework emphasizes a Safety II first approach while keeping Safety I aspects for emergency measures, introduces cognitive and psychophysiological models and utilizes AI enabled decision making in combination with a human-in-the-loop approach. The paper concludes with a comparison on the distribution of Safety II versus Safety I in industrial assistance systems, a list of the main advantages and challenges of Safety II in such systems and a list of recommendations for embedding Safety II in future assistance systems.