Journal
NEURAL NETWORKS
Volume 14, Issue 8, Pages 1099-1112Publisher
PERGAMON-ELSEVIER SCIENCE LTD
DOI: 10.1016/S0893-6080(01)00048-X
Keywords
Intelligent Vehicle navigation; Pseudo Outer-Product Fuzzy Neural Networks; POPFNN; POPFNN-CRI; novel automated pilot system; fuzzy set and rule base; velocity control; automatic transmission; fuzzy set and rule learning; fuzzy control systems
Ask authors/readers for more resources
This paper presents part of our research work concerned with the realisation of an Intelligent Vehicle and the technologies required for its routing, navigation, and control. An automated driver prototype has been developed using a self-organising fuzzy rule-based system (POPFNN-CRI(S)) to model and subsequently emulate human driving expertise, The ability of fuzzy logic to represent vague information using linguistic variables makes it a powerful tool to develop rule-based control systems when an exact working model is not available, as is the case of any vehicle-driving task. Designing a fuzzy system, however, is a complex endeavour, due to the need to define the variables and their associated fuzzy sets, and determine a suitable rule base. Many efforts have thus been devoted to automating this process, yielding the development of learning and optimisation techniques. One of them is the family of POP-FNNs, or Pseudo-Outer Product Fuzzy Neural Networks (TVR, AARS(S), AARS(NS), CRI, Yager). These generic self-organising neural networks developed at the Intelligent Systems Laboratory (ISL/NTU) are based on formal fuzzy mathematical theory and are able to objectively extract a fuzzy rule base from training data. In this application, a driving simulator has been developed, that integrates a detailed model of the car dynamics, complete with engine characteristics and environmental parameters, and an OpenGL-based 3D-simulation interface coupled with driving wheel and accelerator/brake pedals. The simulator has been used on various road scenarios to record from a human pilot driving data consisting of steering and speed control actions associated to road features. Specifically, the POPFNN-CRI(S) system is used to cluster the data and extract a fuzzy rule base modelling the human driving behaviour. Finally, the effectiveness of the generated rule base has been validated using the simulator in autopilot mode. (C) 2001 Elsevier Science Ltd. All rights reserved.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available