Digital Twin Analysis for Driving Risks Based on Virtual Physical Simulation Technology

Abstract

A digital twin is a mapping of real world objects in virtual space. For the study of traffic safety issues, digital twins have great potential to facilitate more accurate and detailed risk analysis. In this study, a digital twin method for highway driving safety analysis is proposed, which consists of three parts: Extracting vehicle motion information in the real world, constructing vehicle motion scenes in the virtual world, and analyzing vehicle driving risks. Firstly, aerial video of vehicle motion is captured by drone, while the microscopic vehicle trajectories are extracted from the video using machine vision algorithms. Secondly, the digital twin of vehicles and roads is constructed, while the motion behavior of vehicles is mapped in a digital space based on virtual physical simulation technology. Finally, according to the stability and trajectory deviation, the driving risks of the vehicle are evaluated, including sideslip, rollover, and guardrail collision. Through a case study, the effectiveness of the proposed digital twin method in driving risk assessment is verified, and one vehicle is found to have a higher driving risk.