RoadGen: Generating Road Scenarios for Autonomous Vehicle Testing
Fan Yang, You Lu, Bihuan Chen, Peng Qin, Xin Peng
TL;DR
RoadGen tackles the lack of diversity in road topology and geometry for autonomous vehicle testing by introducing eight parameterized road components and a guided connection algorithm to build diverse road networks. It further enforces topology diversity with a deduplication metric and delivers ready-to-use HD maps and 3D scenes for joint simulation. Experimental results show RoadGen outperforms random generation in both diversity and speed, and over 92% of sampled road scenarios are usable for joint simulation, demonstrating practical utility. The work also provides a public dataset of road scenarios to facilitate broader simulation-based testing and benchmarking.
Abstract
With the rapid development of autonomous vehicles, there is an increasing demand for scenario-based testing to simulate diverse driving scenarios. However, as the base of any driving scenarios, road scenarios (e.g., road topology and geometry) have received little attention by the literature. Despite several advances, they either generate basic road components without a complete road network, or generate a complete road network but with simple road components. The resulting road scenarios lack diversity in both topology and geometry. To address this problem, we propose RoadGen to systematically generate diverse road scenarios. The key idea is to connect eight types of parameterized road components to form road scenarios with high diversity in topology and geometry. Our evaluation has demonstrated the effectiveness and usefulness of RoadGen in generating diverse road scenarios for simulation.