Assessing the Quality of Concrete Scenarios Generated from Scenario Classes for Automated Driving

Key points

• Once a set of scenario classes covering the operational design domain is available, concrete scenarios are generated from these scenario classes, e.g., to be used for hazard analysis and risk assessment or for the system's verification and validation.1
• In this work, we aggregate scenario distance measures into so-called quality functions that quantify the diversity of concrete scenario collections in the context of automated driving.1
• In particular, we consider the case where scenario classes are formalized as constraint systems and the associated concrete scenario space is not explicitly known.1
• To show the practical relevance of quality functions, we perform a variety of experiments in the context of an abstract scenario-based virtual verification toolchain for automated driving systems.1

Sources (1)

• [1]Assessing the Quality of Concrete Scenarios Generated from Scenario Classes for Automated Driving
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 11:15 AM

  A scenario-based development approach has become state of the art for automated driving systems.

  Once a set of scenario classes covering the operational design domain is available, concrete scenarios are generated from these scenario classes, e.g., to be used for hazard analysis and risk assessment or for the system's verification and validation.

Extractive summary: sentences quoted from the sources.