A Case Study on Formally Validating Motion Rules for Autonomous Cars - Archive ouverte HAL Access content directly
Conference Papers Year : 2021

A Case Study on Formally Validating Motion Rules for Autonomous Cars

Abstract

Car motion control is a key functional stage for providing advanced assisted or autonomous driving capabilities to vehicles. Car motion is subject to strict safety rules which are normally expressed in natural language. As such, these natural language rules are subject to potential misinterpretation during the implementation phase of the motion control stage. In this paper, we show a novel approach by which safety rules are expressed in natural language, then in a formal language specification which is then validated and used to generate a car motion checker. We present a case study of using the approach with true road capture data and its associated imperfections. We also show how the approach lowers the validation efforts needed to guarantee that the car motion always respects a desired set of safety rules while other traditional validation methods would be much heavier to deploy and error prone.
Fichier principal
Vignette du fichier
Formally_validating_motion_rules_for_autonomous_cars.pdf (1.34 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-03331230 , version 1 (09-09-2021)

Identifiers

  • HAL Id : hal-03331230 , version 1

Cite

Mario Henrique Cruz Torres, Jean-Pierre Giacalone, Joelle Abou Faysal. A Case Study on Formally Validating Motion Rules for Autonomous Cars. SEFM 2020 - Collocated Workshops - Software Engineering and Formal Methods, Sep 2020, Amsterdam, Netherlands. pp.233-248. ⟨hal-03331230⟩
81 View
91 Download

Share

Gmail Facebook Twitter LinkedIn More