Marco Trincavelli and Amy Loutfi
An inspection of signal dynamics using an open sampling system for gas identification
Workshop in Networked and Mobile Robot Olfaction in Natural, Dynamic Environments, May 8, 2010, Anchorage, Alaska.
Abstract: Odour discrimination with a sensor array directly exposed to the environment is of great interest in many applications ranging from environmental monitoring to search and rescue and exploration of hazardous areas. Metal oxide based sensors are typically used in such applications for gas detection as they are compact, low costing and exhibit fast response time to an analyte. Given the characteristics of the metal oxide sensors as well as the chaotic nature of the airflow in a natural environment, it has been assumed that a reliable odour discrimination system working in this condition has to work with features that capture the transient characteristics of the signal. In this work we present an experimental setup that enables a deeper insight into transient-based analysis for open sampling systems. Observations on the properties of the signal collected under a controlled experimental condition using an open sampling system are presented. These observations suggest that in a scenario in which the sensors are exposed to different compounds without the possibility to recover the baseline value in between a gas identification, a model of the dynamics of the system is needed and a notion of the sensor state should be maintained that captures the past history of the sensor response.
Keywords: Mobile Robotics Olfaction, Gas Discrimination, Feature Selection.
Download:
Paper: [PDF (2.5MB)]
@INPROCEEDINGS{Trincavelli_Loutfi:ICRAWS:2010,
  AUTHOR = {Trincavelli, Marco and Loutfi, Amy},
  TITLE = {An inspection of signal dynamics using an open sampling system for gas identification},
  BOOKTITLE = {Workshop in Networked and Mobile Robot Olfaction in Natural, Dynamic Environments, ICRA},
  YEAR = {2010},
  DATE = {May 3--8},
  ADRESS = {Anchorage, Alaska, USA},
}