I. Introduction
High-definition (HD) maps consist of roads, lanes, signs, buildings, and other road objects, essential in creating a realistic world in a simulator for autonomous vehicles (AV). Using HD maps in simulation-based testing improves the generalizability of test results to the real world because HD maps are detailed, accurate, extensive, and photo-realistic, thereby permitting an AV to use multiple sensors to build a 3D world model. An essential aspect of HD maps is road geometries’ overall realism and variation. Due to their complexity, roundabouts are often omitted from procedural road network generators, yet they do occur in the real world. In some cities, they are very common. Since there are roundabouts of various shapes and sizes [1], an AV needs to have access to a wide variety of roundabout shapes for thorough simulation-based testing.