Synthesizing Traffic Scenarios from Formal Specifications Using Reachability Analysis

Scenario-based testing is a promising approach for the verification of automated vehicles (AVs). In this paper, we present a novel approach that combines reachability analysis and numerical optimization to derive concrete scenarios from formal specifications. This promises to address multiple deficiencies of previous approaches: Improvement of computation times, handling of nonlinear specifications (e.g., traffic rules), and incorporation of criticality metrics. Our evaluation shows that the computation time increases linearly with the number of agents and the time horizon, compared to a typically exponential increase for methods without reachability analysis.