In the 1980s work began on a new instrument, mimicking some concepts of the mammalian olfactory system (Persaud and Dodd 1982). Since then, these types of devices have been investigated in depth and developed for many applications, such as in the food and pharmaceutical industries, in environmental control, clinical diagnostics, and more (Pearce et al. 2003). The rationale behind the functioning of these instruments has to be found in the use of a few different but relatively nonselective chemical sensors. Their responses are not directly correlated to a specific chemical substance, but rather to the whole repertoire of chemical information contained in an odor, which is often a mixture of many substances. This somewhat resembles the way the biological olfactory system deals with odorants. However, biological olfaction outperforms current chemical analysis instrumentation in specificity, response time, detection limits, coding capacity, stability over time, robustness, size, power consumption, and portability. This is mainly due to the unique architecture of the olfactory pathway, characterized by a high level of sensor redundancy, early combinatorial coding, and exceptional information processing mechanisms.
The fabrication of instrumentation having similar performance and efficiency to those found in biological systems was impeded by the lack of knowledge of the functional anatomy and biochemistry of the olfactory system and the odor information processing pathway. This scenario changed during the last few decades. Several experimental findings have improved the understanding of the olfactory structure (Haberly 2001; Yokoi et al. 1995; Ressler et al. 1994; Shepherd 1987, 1994; Vassar et al. 1994; Buck and Axel 1991; Haberly and Bower 1989) and coding mechanisms inspiring new solutions that enhance the capability of artificial olfaction systems.
In this context, a European project, NEUROCHEM, was funded with the aim of developing a novel computing paradigm for chemical sensing based on the information processing performed in the olfactory pathways in the brain, and also of building a large chemical sensor array for generating realistic data inputs for the models. This chapter describes the part of the outcome of the NEUROCHEM project focused on developing and testing the sensor array. The research planned to fabricate a huge number of chemical sensors, 65,536, of tens of different types, exhibiting broad and overlapping sensitivities and different dynamic ranges. Hence, the array has to implement key features of the biological system, as the receptor redundancy, which are expected to contribute to the achievement of an artificial olfactory device having unique performance and able to generate useful data at a speed comparable to biological olfaction to feed into computational models.
© 2013 by Taylor & Francis Group, LLC.