Latin American and Caribbean Consortium of Engineering Institutions

Conference Track:  Emerging Technologies

Selection1:       Paper

Language:         English

Keywords:         Intelligenet control, Fuzzy Control, Neuro-control

Contact Title:    Prof.

Contact First Name: Antonio

Contact Last Name:  Moran

University:       Universidad Peruana de Ciencias Aplicadas UPC

Web:              http://www.upc.edu.pe

Position:         Director of Electrical Engineering Department

Country:          PERU

Email:            amoran@upc.edu.pe

Fax:              (51) 1 3133344

Paper Title:

Intelligent Car Parking Using Fuzzy-Neural Networks

Abstract:  

This project is concerned with the design of a control system for the autonomous positioning of vehicles in a parking lot. Three control structures have been proposed and analyzed: nonlinear LQ controller, fuzzy controller and fuzzy-neural controller, being the last one which has the better performance.

The fuzzy-neural control system has been designed in two steps. First, a fuzzy controller was constructed using the experience of a human driver capable to park the vehicle in the desired position. Next, a multilayer neural network is configured to represent the structure of the fuzzy controller with the same input-output characteristics. At this step, both fuzzy controller and fuzzy-neural controller perform almost the same. Then the neural network starts its learning process to minimize a cost function defined in terms of the positioning error. After several steps of learning, the fuzzy-neural network presents better results than the fuzzy controller, with improved positioning accuracy and obstacle avoidance capabilities.

The fuzzy-neural network has been trained considering the dynamics of a low speed vehicle. To do that, a modified structure of the Dynamic Back Propagation algorithm was proposed to optimize computing time as well as memory requirements.

The effectiveness of the fuzzy-neural control system was verified by computer simulations and experimental analysis using a small car-like mobile robot. The results show that the vehicle is able to achieve the desired parking position avoiding obstacles placed closely to the desired position. Also the vehicle is able to follow any desired trajectory starting from arbitrary initial positions.

Mailing Address:

Universidad Peruana de Ciencias Aplicadas UPC

Av. Prolongacion Primavera 2390,

Monterrico, Surco

Lima

PERU

Phone:

(51) 1 3133333

(51) 1 4483340

Authors:  

(1) Antonio Moran,  amoran@upc.edu.pe

Universidad Peruana de Ciendas Aplicadas UPC, (P)

(2) Minoru Hayase, Tokyo University of Agric. and Technology, Japan