Vehicle collision avoidance system using embedded hybrid intelligent prediction based on vision/GPS sensing

Chung-Ping Young, Bao Rong Chang, Hsiu Fen Tsai, Ren Yang Fang, Jian Jr Lin

Research output: Contribution to journalArticlepeer-review

7 Citations (Scopus)

Abstract

Lane marking detection assists drivers to judge any unattended deviation on the roadway. Thus, instead of radar or laser sensor, vision/GPS sensing has been introduced not only to recognize lane marking ahead, but also to detect vehicle in front or around while driving, particularly enabling the collision warning or deviation alert. In this paper, a high-performance collision avoidance system (CAS) has realized vision/GPS sensing for active vehicle safety where GPS supplying the localization information to col-laborate with vision scanning in a very short time, judges whether or not an impending crash may be caused. Besides, in order to retrieve the information from heterogeneous data collected by both out-of-vehicle and in-vehicle sensing, a two-layer embedded data fusion, which is a quantum-tuned back-propagation neural network (QT-BPNN) plus anadaptive network-based fuzzy inference system (ANFIS), has implemented in a distributed dual-platform DaVinci+XScale-NAV270 for achieving a fast response to collision warning and event data recording. Finally, based on performance evaluation the experiments showed that the proposed method outperforms two well-known alternative systems, AWS-Mobileye and ACWS-Delphi.

Original languageEnglish
Pages (from-to)4453-4468
Number of pages16
JournalInternational Journal of Innovative Computing, Information and Control
Volume5
Issue number12
Publication statusPublished - 2009 Dec

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Software
  • Information Systems
  • Computational Theory and Mathematics

Fingerprint Dive into the research topics of 'Vehicle collision avoidance system using embedded hybrid intelligent prediction based on vision/GPS sensing'. Together they form a unique fingerprint.

Cite this