딥러닝을 활용한 실가상 데이터 기반가상환경 차량 동역학 모델링 자동화 프로세스 구축

초록

Since various systems are becoming automated, ensuring the safety and robustness of their software is crucial. Automated systems in robots and/or vehicles often rely on various virtual simulation pipelines, as failures observed on hardware platforms can lead to critical safety issues. The introduced simulation system can support virtual testing of the automated system, and digital twin technology is combined to replace the hardware testing procedure. However, the dynamic characteristics of the platform must be modeled using numerous parameters, many of which are difficult to obtain in real- world settings. Therefore, in this study, we introduce an automated pipeline for setting up a virtual hardware platform that takes dynamic characteristics into account, using vehicle maneuvering datasets from both real-world and virtual environments. This deep learning-based automated virtual dynamics setup pipeline ensures sufficient modeling scalability and accuracy to serve as an effective substitute for hardware platform testing.

키워드