Autonomous Driving System : Model Based Safety Analysis

Mohamed Tlig (1) , Mathilde Machin (2) , Romain Kerneis (1) , Emmanuel Arbaretier (2) , Linda Zhao (3) , Florent Meurville (4) , Jean van Frank (1)

1 IRT SystemX
2 APSYS [Elancourt]
3 SECTOR - SECTOR
4 Valeo Driving Assistance Domain

The desire to introduce autonomous vehicles by 2020 on the roads represents a real technological challenge. It requires breaks with traditional design, security and validation processes to achieve a safe system. As part of the SVA 1 (Simulation of Autonomous Vehicle Safety) project, we present the process under development at the Institute for Technological Research SystemX 2 , in order to optimally address the limitations of existing methods. The objective is to provide designers methods and tools to support safety considerations during the design and the validation phases of autonomous vehicles functions. In this paper, we apply a Model Based Safety Analysis methodology (MBSA) to an Advanced Driver-Assistance System (ADAS), using a modular numerical simulation platform. We describe the different activities carried out during each stage and define the associated objectives. Experimental simulation results are presented, showing the advantages of such approach.

Mots clés: AltaRica; Functional Safety; Automated Driving Systems; MBSA; ISO 26262; Autonomous critical systems

Domaines: Intelligence artificielle [cs.AI]

Mohamed Tlig, Mathilde Machin, Romain Kerneis, Emmanuel Arbaretier, Linda Zhao, et al.. Autonomous Driving System : Model Based Safety Analysis. 2018 48th Annual IEEE/IFIP International Conference on Dependable Systems and Networks Workshops (DSN-W), Jun 2018, Luxembourg city, Luxembourg. ⟨10.1109/DSN-W.2018.00012⟩. ⟨hal-01906465⟩