Early Model-Based Safety Analysis for Collaborative Robotic Systems

The current era is marked by an accelerated digitization of manufacturing processes, with robotic systems increasingly integrated into various workflows. Yet, despite significant advancements, it is impractical to fully automate certain tasks due to prohibitive costs and technical constraints. As a result, there’s a growing emphasis on human-robot collaboration (HRC) for intricate operations. In HRC scenarios, humans and robots co-inhabit the same work environment, operating side by side. More than just mere coexistence in the same space, they actively collaborate on shared tasks, thus raising the stakes in terms of safety. The dynamic behavior of robots must be synchronized with the anticipated and unexpected human actions, adding another layer of complexity to the safety considerations. It is essential to conduct comprehensive safety analyses that identify potential risks that pose harm to the human operator. As a proactive measure to foster early-stage safety and risk analysis, we propose the use of goal models. The approach enables the specification of safety threats within the HRC context, thereby facilitating the development of safety tasks and supportive monitoring mechanisms. This approach helps in the refinement and implementation of safety measures, ensuring a secure and productive environment for human-robot collaboration. Note to Practitioners—This paper was motivated by the need for conducting thorough safety analyses in future manufacturing scenarios. This is particularly the case as the use of HRC becomes more common in industry. This is fostered by the ideas of smart manufacturing and the spread of collaborative robots (cobots) in work environments. To support proper safety analyses in early phases of development, we propose a model-based methodology to relate the requirements of the system with the corresponding safety hazards and its countermeasures. The proposed model-based approach places particular emphasis on the identification of hazards not only related to a single system but resulting from collaboration scenarios. Thus, it contributes to an extensive safety analysis of collaborative robotic systems right in the early development stages. Application to a manufacturing case example showed its applicability and usefulness to identify potential safety hazards early in development. Results, however, also indicate that future work is needed for implementing a seamless view concept allowing the derivation of models representing specific situations to be closely investigated. For this, abstraction mechanism can aid in guiding the safety analysis procedure.