Sign In

Communications of the ACM

ACM TechNews

Turing Tests and the Problem of Artificial Olfaction

View as: Print Mobile App Share:
The physical components of human olfaction.

David Harel is working to develop a means of reproducing scent artificially.


The ability to reproduce scent artificially, which involves measuring an odor at one point in space and then replicating it in another, is surprisingly complex, and the work of the Weizmann Institute of Science's David Harel sheds light on the issue.

Harel says smell reproduction consists of three components: a "sniffer" device that converts an input odor into a digital signature, a "whiffer" device containing a spectrum of fixed smells that can be combined and issued in carefully measured quantities and concentrations, and the sniffer/whiffer interface. "[This] analyzes the signature coming from the sniffer and instructs the whiffer as to how it should mix its pallet odorants to produce an output odor that is perceived by a human to be as close as possible to the original input," Harel says.

However, he notes that unlike with images or sounds, approximation of odors is not good enough.

Using the Turing test as a rough template, Harel has developed an approach in which a human is asked to differentiate real odors from those generated by an artificial olfactory system. Via immersion in audio and video of the locale where the smell was collected, the subject can decide whether the related odor is real or artificial, with the system's performance gauged by repeating the procedure with many different samples and testers.

From Technology Review
View Full Article


Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA


No entries found

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account