Development of Urban Space Environment for Driving Simulation of Autonomous Vehicles and Electric Vehicles

Abstract

In this study, we developed an urban space environment for driving simulation of autonomous and electric vehicles at low cost using open data and a game engine. In the development of the control algorithm of the automobile, the trial frequency is limited, because the test using the actual car is expensive. And, it is difficult to freely set the change of road conditions by weather and time. Therefore, the development using the simulation by the virtual space is carried out. Since these are carried out by the calculation in the computer, there is no limit on various condition changes and trial frequency. On the other hand, it is necessary to carry out tests in an environment suitable for actual environment in order to carry out control such as obtaining information from actual environment, because autonomous vehicles and electric vehicles use sensors such as LIDAR to obtain information from the real environment and to calculate efficiency from road data such as slope and road surface conditions. However, high cost is required to construct the environment which perfectly imitates the real environment in the virtual environment. In this study, a large-scale urban space model with information necessary for a driving simulator was developed by combining open data of Geospatial Information Authority of Japan with inexpensive development environments such as game engines.