Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Concretization of Abstract Traffic Scene Specifications Using Metaheuristic Search

Aren A. Babikian, Oszkár Semeráth, Dániel Varró

TL;DR

This work tackles automated concretization of abstract traffic scene specifications into numeric, simulation-ready initial scenes for autonomous-vehicle testing. It introduces a 4-valued partial-model functional scene specification language, a formal mapping from abstract relations to numeric constraints, and a metaheuristic search framework to derive concrete scenes on arbitrary maps. The authors prove soundness of the numeric-to-functional refinement and demonstrate that their MHS-based approach achieves higher success rates and scalability than the Scenic baseline across multiple road maps, at the cost of longer runtimes for larger problems. Practically, the method supports rigorous scenario-based testing and paves the way for more scalable AV safety certification, including integration with vision-based ML component testing in CARLA.

Abstract

Existing safety assurance approaches for autonomous vehicles (AVs) perform system-level safety evaluation by placing the AV-under-test in challenging traffic scenarios captured by abstract scenario specifications and investigated in realistic traffic simulators. As a first step towards scenario-based testing of AVs, the initial scene of a traffic scenario must be concretized. In this context, the scene concretization challenge takes as input a high-level specification of abstract traffic scenes and aims to map them to concrete scenes where exact numeric initial values are defined for each attribute of a vehicle (e.g. position or velocity). In this paper, we propose a traffic scene concretization approach that places vehicles on realistic road maps such that they satisfy an extensible set of abstract constraints defined by an expressive scene specification language which also supports static detection of inconsistencies. Then, abstract constraints are mapped to corresponding numeric constraints, which are solved by metaheuristic search with customizable objective functions and constraint aggregation strategies. We conduct a series of experiments over three realistic road maps to compare eight configurations of our approach with three variations of the state-of-the-art Scenic tool, and to evaluate its scalability.

Concretization of Abstract Traffic Scene Specifications Using Metaheuristic Search

TL;DR

This work tackles automated concretization of abstract traffic scene specifications into numeric, simulation-ready initial scenes for autonomous-vehicle testing. It introduces a 4-valued partial-model functional scene specification language, a formal mapping from abstract relations to numeric constraints, and a metaheuristic search framework to derive concrete scenes on arbitrary maps. The authors prove soundness of the numeric-to-functional refinement and demonstrate that their MHS-based approach achieves higher success rates and scalability than the Scenic baseline across multiple road maps, at the cost of longer runtimes for larger problems. Practically, the method supports rigorous scenario-based testing and paves the way for more scalable AV safety certification, including integration with vision-based ML component testing in CARLA.

Abstract

Existing safety assurance approaches for autonomous vehicles (AVs) perform system-level safety evaluation by placing the AV-under-test in challenging traffic scenarios captured by abstract scenario specifications and investigated in realistic traffic simulators. As a first step towards scenario-based testing of AVs, the initial scene of a traffic scenario must be concretized. In this context, the scene concretization challenge takes as input a high-level specification of abstract traffic scenes and aims to map them to concrete scenes where exact numeric initial values are defined for each attribute of a vehicle (e.g. position or velocity). In this paper, we propose a traffic scene concretization approach that places vehicles on realistic road maps such that they satisfy an extensible set of abstract constraints defined by an expressive scene specification language which also supports static detection of inconsistencies. Then, abstract constraints are mapped to corresponding numeric constraints, which are solved by metaheuristic search with customizable objective functions and constraint aggregation strategies. We conduct a series of experiments over three realistic road maps to compare eight configurations of our approach with three variations of the state-of-the-art Scenic tool, and to evaluate its scalability.
Paper Structure (31 sections, 2 equations, 13 figures, 3 tables)

This paper contains 31 sections, 2 equations, 13 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Preliminaries
  3. Traffic scenes and scenarios
  4. Levels of abstraction in traffic scenarios
  5. Scenario-based testing by simulation
  6. Scene concretization in scenario-based testing
  7. Functional scene specification
  8. Scene specification language
  9. Static analysis of scene specifications
  10. Logical scenes as numeric problems
  11. Numeric concretization problem
  12. Mapping functional relations to logical constraints
  13. Benefits of the mapping
  14. Using MHS to derive concrete scenes
  15. Evaluation
  16. ...and 16 more sections

Figures (13)

  • Figure 1: A traffic scene involving three vehicles
  • Figure 2: An initial traffic scene described at various levels of abstraction
  • Figure 3: A functional-level scene specification
  • Figure 4: Concrete scene that does not satisfy \ref{['fig:func-full']}, but satisfies the corresponding Scenic-expressible subset
  • Figure 5: Mapping from functional relations to logical constraints
  • ...and 8 more figures

Theorems & Definitions (2)

  • proof
  • proof