Int. J. of Industrial and Systems Engineering   »   2008 Vol.3, No.1

 

 

Title: Study on modelling obstacles-avoidance behaviour of virtual driver based on multilayer fuzzy neural network

 

Author: Lou Yan, He Hanwu, Zheng Detao, Lu Yongming

 

Addresses:
School of Mechantronics and Control Engineering, Shenzhen University, Nanshan District, Shen Zhen City, Guang Dong Province, P.R.China.
Faculty of Electro-Mechanical Engineering, Guang Dong University of Technology, Guangzhou 510090, PR China.
Faculty of Electro-Mechanical Engineering, Guang Dong University of Technology, Guangzhou 510090, PR China.
Faculty of Electro-Mechanical Engineering, Guang Dong University of Technology, Guangzhou 510090, PR China

 

Abstract: In virtual traffic scene of driving simulator, a fuzzy neural network controller Vision Fuzzy Back Propagation (VFBP) for modelling human-like driving obstacle-avoidance behaviour was developed. The obstacle-avoidance behaviour of virtual driver is inspired by the vision information based on real driving behaviour characteristic. VFBP controller consists of six layers. Considering the influence of the vehicle speed on the fuzzy distance, behaviour confidence and award coefficient are used to optimise fuzzy rules. Assessing level is used to ensure behaviour security. Adventure coefficient is used to call VFBP controller. In addition, VFBP controller can avoid the local smallest problem of the potential method, which makes the virtual vehicle go more smoothly. VFBP controller can simulate individual driver's behaviour through VFBP online learning.

 

Keywords: VFBP controller; obstacle avoidance; avoidance behaviour; individual virtual drivers; behaviour confidence; adventure coefficient; fuzzy neural networks; virtual traffic; driving simulator; modelling; vision information; simulation; collision avoidance.

 

DOI: 10.1504/IJISE.2008.015915

 

Int. J. of Industrial and Systems Engineering, 2008 Vol.3, No.1, pp.70 - 86

 

Available online: 02 Dec 2007

 

 

Editors Full text accessAccess for SubscribersPurchase this articleComment on this article