A Neuro-Fuzzy System for Extracting Environment Features Based on Ultrasonic Sensors

Sensors

Graciliano Nicolás Marichal¹  Angela Hernández²  Leopoldo Acosta² 

[1] Departamento de Ingeniería de Sistemas y Automática, Arquitectura y Tecnología de Computadoras, Universidad de La Laguna, La Laguna, Tenerife 38071, Spain;

关键词: neural networks;    fuzzy systems;    neuro-fuzzy;    ultrasonic sensing;    autonomous robots;   

DOI  :  10.3390/s91210023

来源: mdpi

【 摘 要 】

In this paper, a method to extract features of the environment based on ultrasonic sensors is presented. A 3D model of a set of sonar systems and a workplace has been developed. The target of this approach is to extract in a short time, while the vehicle is moving, features of the environment. Particularly, the approach shown in this paper has been focused on determining walls and corners, which are very common environment features. In order to prove the viability of the devised approach, a 3D simulated environment has been built. A Neuro-Fuzzy strategy has been used in order to extract environment features from this simulated model. Several trials have been carried out, obtaining satisfactory results in this context. After that, some experimental tests have been conducted using a real vehicle with a set of sonar systems. The obtained results reveal the satisfactory generalization properties of the approach in this case.

【 授权许可】

CC BY   
© 2009 by the authors; licensee Molecular Diversity Preservation International, Basel, Switzerland.