A Toolchain for Synthesizing and Validating Safety Architectures

Yuri Gil Dantas, Tiziano Munaro, Carmen Cârlan, Vivek Nigam, Simon Barner, Shiqing Fan, Alexander Pretschner, Ulrich Schöpp and Sergey Tverdyshev

To appear in SN Computer Science,

2023

abstract

Autonomous vehicles handle complicated tasks that may lead to harm when performed incorrectly. These harms, in particular when caused by system faults, may be avoided by the deployment of safety architectural patterns, such as the Heterogeneous Duplex pattern. Our goal is to provide safety engineers with computer-aided support for synthesizing architectures with safety architecture patterns. To this end, we build on our previous work in which we proposed a model-based system engineering plugin to enable the model-driven approach using safety architecture patterns. This article proposes a toolchain for synthesizing the structure and switching logic of safety architectures, as well as for validating them through simulation-based fault-injection. We validate our toolchain using an industrial use-case for autonomous driving systems, namely a Highway Pilot system.

subject terms: Model-based systems engineering, Toolchain, Safety architecture patterns, Reconfiguration, Simulation, MbSE, AutoFOCUS3, DSE