Recognition of Humaa Behaviors by Mahalanobis'Generalized Distance in Odor-sensor Characteristics

• Sawada Ayako

  Kanazawa University of Economics

• Oyabu Takashi

  Kanazawa University of Economics

• Nanto Hidehito

  Kanazawa Institute of Technology

Bibliographic Information

Other Title

• ニオイセンサのマハラノビスの距離による人間のふるまい判別

Abstract

Human beings generate many kinds of odors in a domicile. The kinds and concentrations of these odors are different according to human behaviors. The odor-sensor characteristics were analyzed for recognition of basic human behaviors in a domicile. Taking meals, sleeping and opening the door were chosen as basic behaviors. The odor sensors are installed at various points in an experimental house. All the odor sensors are a metal oxide type. Human behaviors were identified by applying the odor sensor output at an installing point to Mahalanobis' generalized distance(MGD). MGD's of sensor outputs for the behaviors were also adapted to Poisson distribution. A simulation was done using the sensor outputs of a different day to evaluate recognition of behaviors. As for the results, it was possible to recognize the behaviors at high probabilities, namely taking meals at 89%, and sleeping at88%.

Journal

• Journal of Life Support Engineering

  Journal of Life Support Engineering 13 (2), 20-25, 2001

  The Society of Life Support Engineering

Keywords

• odor sensor, odor sensor pattern, welfare, Mahalanobis' generalized distance

Details

• CRID: 1390282679343400960
• NII Article ID: 10013402509
• NII Book ID: AA11947931
• DOI: 10.5136/lifesupport.13.20
• ISSN: 18845827
• ISSN: 13419455
• Text Lang: ja
• Data Source:
  • JaLC
  • Crossref
  • CiNii Articles
• Abstract License Flag: Disallowed