Advancing Safe Robot Behavior by Formalizing STPA with Tumato

Abstract

As autonomous robotic systems gain popularity, ensuring their safety and reliability becomes increasingly important. Inadequate attention to safety can lead to severe consequences, especially when operating alongside human agents or in other safety-critical applications. The growing complexity of these systems and their tasks amplifies the challenge of managing safe behavior. Systems Theoretic Process Analysis (STPA) offers a promising approach to system safety. However, current applications of STPA are often manual and lack the necessary formality and automation for rigorous safety. This informality results in inherent ambiguities and fails to guarantee consistency, soundness, and completeness. In this paper, we explore how Tumato 2.0, a constraint-based behavior planning approach, can formalize and automate a substantial part of the STPA process, providing an integrated safety workflow. By linking Tumato's automated approach with STPA's holistic view on safety, we obtain a formalized integration of these methods. We demonstrate this integrated approach through a case study featuring an autonomous mobile robot, showcasing the potential to enhance the safety of autonomous robotic systems.