Abstract
Recently, virtual environment-based techniques to train sensor-based autonomous driving models have been widely employed due to their efficiency. However, a simulated virtual environment is required to be highly similar to its real-world counterpart to ensure the applicability of such models to actual autonomous vehicles. Though advances in hardware and three-dimensional graphics engine technology have enabled the creation of realistic virtual driving environments, the myriad of scenarios occurring in the real world can only be simulated up to a limited extent. In this study, a scenario simulation and modeling framework that simulates the behavior of objects that may be encountered while driving is proposed to address this problem. This framework maximizes the number of scenarios, their types, and the driving experience in a virtual environment. Furthermore, a simulator was implemented and employed to evaluate the performance of the proposed framework.
Highlights
We focus on virtual scenario simulation for autonomous driving simulators rather than developing technologies for autonomous vehicles or their physical dynamics
This paper proposes a virtual scenario simulation and modeling framework for autonomous driving simulators that naturally and automatically creates both normal and hazardous scenarios, as well as providing a facile graphical user interface (GUI) to non-developer users to edit scenarios before and during simulation and modify season, weather, and lighting conditions
We propose a virtual scenario simulation and modeling framework for simulators used to train and evaluate autonomous vehicles
Summary
The training of an unmanned ground vehicle using learning-based algorithms requires the collection of a significant amount of data outdoors [1], which is constrained by weather, time, and considerations for the safety of other vehicles and pedestrians To resolve such problems, simulation technology based on virtual environments has been widely employed in skill training [2,3,4,5,6,7]. To ensure the robust training and accurate evaluation of autonomous driving algorithms, deployed virtual environments should include realistic agents like vehicles, humans, and animals, in addition to fundamental driving environment conditions They should be capable of generating a wide variety of natural scenarios—in addition to normal scenarios wherein all agents obey traffic rules, hazardous scenarios that may induce accidents should be simulated.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call