Tiered Model-Based Safety Assessment

Abstract

Processes and techniques used for assessing the safety of a complex system are well-addressed by safety standards. These standards usually recommend to decompose the assessment process into different stages of analysis, so called tiered safety assessment. Each analysis stage should be performed by applying recommended assessment techniques. To provide confidence in the correctness of the whole analysis, some verification techniques, usually traceability checking, are applied between two stages. Even if the traceability provides some confidence in the correctness of the decomposition, the following problems remains How to model the system behaviours at each stage of safety assessment? How to efficiently use these stages during the design process? What is the formal relationship between these modelling stages? To tackle these problems, we propose a way to specify, formalize and implement the relations between assessment stages. The proposal and its pros & cons are illustrated on a Remotely Piloted Aircraft System (RPAS) use-case.

DROSERA: A DROne Simulation Environment for Risk Assessment

“…In this work, engine failure is considered as the most probable event leading to a loss of control on the trajectory according to returns on experience from manufacturers and literature. It can also be computed for example using Model Based Safety Assessment methods considering the whole physical and software architectures of the UAV (Bieber et al (2018), Delmas et al (2019)). The second term in (1) refers to the probability of impact at ground and should be evaluated for all possible impact locations of the UAV.…”

SafeSens - Uncertainty Quantification of Complex Perception Systems

“…Similarly in [14] the authors state that safety assessment (including validation) shall be performed throughout different design stages. Their main contribution is a proposal for a novel formalization of relations between safety assessments.…”

“…an approximation of a specific DNN, can easily be integrated. Also, as with [14], no uncertainty quantification is part of the final result.…”

ECMS: Energy-Efficient Collaborative Multi-UAV Surveillance System for Inaccessible Regions

“…The steady flight power consumption model consists of power for lift, power to overcome drag, and ascending/descending power. In literature, we find drone systems can use more than one onboard power supply for a more reliable operation [23], [24].…”