This publication documents the algorithmic core of the AVSE Real-Time Simulation Miner, a tool developed to detect and extract simulated driving scenarios during execution in the CARLA simulator, based on logical scenario descriptions from the MOSAR Scenario Manager. The document provides a detailed technical specification of the end-to-end pipeline that links MOSAR logical scenarios (road infrastructure and kinematic constraints) to dynamic simulation traces. It is structured around three coupled problems solved online at each simulation tick in a neighbourhood of a chosen ego actor: Scenario detection — initialization and operation of a LogicalScenarioDetector, including infrastructure consistency checks, exploration of candidate actor mappings, and temporal validation of multi-step scenario instances. Infrastructure handling — construction and transformation of comparable scenario and simulation graphs from CARLA map topology (segment merging, road extremity nodes, bidirectional connectivity, relative connection sides, chain merging, and longitudinal position alignment). Actor handling — formalization of kinematic constraints, construction of candidate forests, and enumeration of globally consistent mappings between scenario and simulation actors using backtracking procedures. This work was carried out in the context of the SYNERGIES project (Horizon Europe), under Task 3.4 — Data generation from advanced simulation tools, and supports data-generation activities contributing to deliverable D3.3 — Data generation report. The publication is intended for researchers and engineers working on connected, cooperative and automated mobility (CCAM), scenario-based testing, and simulation-driven dataset generation.
Laurent et al. (Mon,) studied this question.