Application of neural networks in vehicle simulation as a substitute for driver models

Koller, Tamás and Tóth-Nagy, Csaba (2025) Application of neural networks in vehicle simulation as a substitute for driver models. COGNITIVE SUSTAINABILITY, 4 (2). ISSN 2939-5240

Preview

Text 20502-ArticleText-81613-1-10-20250904.pdf - Published Version Available under License Creative Commons Attribution Non-commercial Share Alike. Download (705kB) | Preview

Abstract

The development and optimization of vehicle simulation models is essential for the virtual validation of new features during vehicle development. New challenges are emerging that require the application and use of innovative solutions. The use and development of artificial intelligence methods can accelerate development processes, which will require a broader investigation of their feasibility. This paper explores the potential of applying a neural network based technology to a driver model within a vehicle simulation instead of the traditional proportional-integral (PI) control methods. The artificial neural network applied here can learn the driving style of the driver and can be used in both simulation and virtual testing scenarios. The aim of this paper is to demonstrate the use of neural network to replace the PI controller throttle signal in a vehicle simulation driver model. In this novel approach, the artificial neural network can learn real driver behavior resulting in a more realistic driver model in vehicle simulation further advancing the accuracy of the simulation.

Actions (login required)

Edit Item