Modeling Safe Adaptation Spaces for Self-Adaptive Systems Using Contextual Safety Concept Trees
SHORT
This program is tentative and subject to change.
Safety-critical autonomous systems operating in complex real-world environments face significant challenges in consistently meeting functional and non-functional requirements. While self-adaptive systems have demonstrated effectiveness in uncertain environments, implementing self-reconfiguration within an adaptation space introduces safety concerns, as the verification of safety in self-adaptive systems remains an unresolved research challenge. In this paper, we propose a novel method for modeling the adaptation space of a self-adaptive system utilizing contextual safety concept trees. Our proposed approach facilitates both design time safety assessment and run-time determination of the subspace of safe adaptations, based on context and system state observations. To address uncertainty in observations, we employ fuzzy inference systems to model context constraints, thereby aggregating imprecise information from multiple sources. The resulting analysis yields a safe adaptation space that can be explored without restrictions in subsequent phases of the adaptation loop. We validate our proposal through a case study in the domain of mobile robotics, demonstrating the suitability of our method for modeling safe adaptation spaces.