DriveSceneGen: Generating Diverse and Realistic Driving Scenarios from Scratch

Shuo Sun, Zekai Gu, Tianchen Sun, Jiawei Sun, Chengran Yuan, Yuhang Han, Dongen Li,
and Marcelo H. Ang Jr.
Advanced Robotics Centre, National University of Singapore
Indicates Equal Contribution
MY ALT TEXT

An illustration of a set of generated scenarios and its components at the various stages of DriveSceneGen pipeline.

Abstract

Realistic and diverse traffic scenarios in large quantities are crucial for the development and validation of autonomous driving systems. However, owing to numerous difficulties in the data collection process and the reliance on intensive annotations, real-world datasets lack sufficient quantity and diversity to support the increasing demand for data. This work introduces DriveSceneGen, a data-driven driving scenario generation method that learns from the real-world driving dataset and generates entire dynamic driving scenarios from scratch. DriveSceneGen is able to generate novel driving scenarios that align with real-world data distributions with high fidelity and diversity. Experimental results on 5k generated scenarios highlight the generation quality, diversity, and scalability compared to real-world datasets. To the best of our knowledge, DriveSceneGen is the first method that generates novel driving scenarios involving both static map elements and dynamic traffic participants from scratch.

Demo Video

BibTeX


        @misc{sun2023drivescenegen,
          title={DriveSceneGen: Generating Diverse and Realistic Driving Scenarios from Scratch}, 
          author={Shuo Sun and Zekai Gu and Tianchen Sun and Jiawei Sun and Chengran Yuan and Yuhang Han and Dongen Li and Marcelo H. Ang Jr au2},
          year={2023},
          eprint={2309.14685},
          archivePrefix={arXiv},
          primaryClass={cs.RO}
        }