A Runtime Safety Analysis Concept for Open Adaptive Systems

Abstract

In the automotive industry, modern cyber-physical systems feature cooperation and autonomy. Such systems share information to enable collaborative functions, allowing dynamic component integration and architecture reconfiguration. Given the safety-critical nature of the applications involved, an approach for addressing safety in the context of reconfiguration impacting functional and non-functional properties at runtime is needed. In this paper, we introduce a concept for runtime safety analysis and decision input for open adaptive systems. We combine static safety analysis and evidence collected during operation to analyse, reason and provide online recommendations to minimize deviation from a systems safe states. We illustrate our concept via an abstract vehicle platooning system use case.

Review of Graph-Based Hazardous Event Detection Methods for Autonomous Driving Systems

“…In an inherently uncertain driving domain, reasoning the influence of a combination of probabilistic events is the strength of BNs. Quantifying uncertainty is crucial for decision making and is utilised clearly in the work of [69]. The authors present a theoretical BN framework for platoon vehicles to detect hazardous events such as a collision between members and speed violations by comparing safe prior with current speed and inter-distance observations.…”

“…The method in [69] is a good representation of BN reasoning but illustrates a common limitation between methods. Hazardous event probabilities were aggregated to infer system states that reflect both actor collision and speed violation.…”

“…As such, metrics centre around spatial and temporal features. Over half of the methods within this survey used spatial features to either semantically describe the relative location (e.g., front, behind, left, right) in the case of [18] or [18], [19], [21], [22], [35], [57], [58], [69], [70], [73], [74], [79]- [82] Temporal [19], [21], [22], [35], [69], [70], [73], [79]- [82] Behaviour [18], [19], [35], [73], [79]- [82] Spatial [18], [22], [35], [55], [56] Temporal [22], [35], [55], [56] Behaviour [18], [35] Road Type [19], [57], [58], [72] Geometry [19] Weather [18], [19],…”

“…Yet, given the importance of the real-time operation, many real-time theoretical frameworks suggested did not test with real perception sensors or as part of live systems [67], [69]. This is sufficient for proof of concept but lacks rigour for real implementation.…”

SafeML: Safety Monitoring of Machine Learning Classifiers Through Statistical Difference Measures

“…• The outcome of the proposed approach can be used as an input for runtime safety analysis in adaptive systems [28,29] 6.2 Limitations of the proposed method…”