A Script Hook-based Ultra-Low Cost Driving Simulator for Development of Self-Driving Algorithms

Abstract

In this paper, we propose a method to build up a low cost driving simulator for the development of Advanced Driving Assistance System (ADAS) and Autonomous Driving (AD) algorithms. The proposed method is based on a low cost physical game engine developed by Rockstar Games, Inc. that is commercially available in the market, thereby cost-effectively available to individuals. In order to implement the ADAS and AD simulator, we present first how to obtain internal data from the game engine and further how to generate useful data for the use of development of ADAS and AD algorithms, e.g. training dataset for deep learning algorithms. In addition, we present a method to generate the simulated sensor data by modeling the actual sensors based on acquired Ground Truth (GT) data. An example of the simulation environment setting through the GUI is presented, and the entire structure of the simulator configuration is also presented.