Image credit (creative commons): Flickr user Bradley Gordon

Digital noses have existed in some form or another for decades. Serious work started on electronic noses as far back as the 1950s, and over the past decade scientists and researchers have made significant gains towards developing systems that replicate aspects of human olfactory perception.

However, despite the progress made on digital noses over the past few decades, the reality is most of the products brought market are nothing like actual noses. In fact, pretty much all electronic sniffers aren’t sniffers at all…at least until now.

That’s because just last week a group of researchers at Brown University unveiled a new system they are calling Trufflebot, an electronic nose that – you guessed it – actually sniffs. As first reported in IEEE Spectrum, the researchers debuted a research paper at a bioengineering conference entitled “TruffleBot: Low-Cost Multi-Parametric Machine Olfaction” that outlines how Trufflebot goes beyond simply using chemical sensors to identify an odor.

From the paper:

…”an important insight into the biological process is that the brain takes advantage of many types of non-chemical information when analyzing odors, including temporal, spatial, mechanical, hedonic, and contextual correlations.”

The reason this non-chemical information is important is that odors consist of more than just chemical reactions replicated in the same way in every situation. As it turns out, odors react to environmental conditions in unique ways. Because of this, the researchers developed a nose that not only detects an odor as determined by a set of chemical sensors, but also factors in changes in environmental conditions when the nose takes a “sniff.”

From the paper:

“The TruffleBot simultaneously samples pressure, temperature, and chemical time series, while “sniffing” in a temporally modulated sequence which introduces spatiotemporal time signatures, such as transport delays and diffusive dynamics. We show how these multidimensional signals depend on chemical and physical properties which can be unique to a particular chemical.”

The Trufflebot System Architecture. Source: Brown University

As can be seen in the figure above, the system is comprised of eight pairs of sensors. Each pair consists of a chemical and a mechanical sensor.  The chemical sensors detect the chemical fingerprint of a vapor, while each mechanical sensor detects changes to air pressure and temperature, important contributors to how we humans process sensory information.

“When animals want to smell something, they don’t just passively expose themselves to the chemical,” lead researcher Jacob Rosenstein told IEEE Spectrum. “They’re actively sniffing for it—sampling the air and moving around—so the signals that are being received are not static.”

By adding additional information, this new approach is also much more accurate according to the Brown researchers. By just using the chemical sensors, the correct odor can be identified 80% of the time. By incorporating the additional information gathered through sniffing in air and understanding the atmospheric conditions of air pressure and temperature, the research team from Brown believes they can get to 95% accuracy.

Despite all this progress, we’re probably still a few years away from affordable pocket electronic noses which means, for now at least, I’ll just have to continue using my own.

Sniff.

Speaking of electronic noses, you can see the Wall Street Journal’s Wilson Rothman put Aryballe’s handheld digital nose to the test live on stage at last year’s Smart Kitchen Summit.  

Leave a Reply