Scentronic Sends Smells Over the Internet With an E-Nose and AI

A photo of freshly brewed coffee cannot carry its aroma. A team of researchers, including scientists at Ohio State University, has built a prototype system called Scentronic that aims to change that by sharing smells over the internet. The work was published in the journal Science Advances.

The system does not transmit scent molecules. On the sending side, an "electronic nose" uses sensors and AI to work out which smell it is detecting. That classification is sent online to a receiving device, which then releases the same type of scent using fragrance materials loaded into it beforehand.

In other words, what travels is information, while the scent-producing material has to be prepared at the other end.

Five scents so far

The prototype recognizes five scents: coffee, cake, orange, beer and red wine. The output devices come in several forms, including a handheld unit and wearable versions worn on the arm or neck.

The approach has clear limits. If the e-nose identifies coffee, the receiving device emits the coffee fragrance it holds, not the particular character of that cup, such as differences between beans or roasts. Reproducing that level of detail remains a challenge. Adding new scents also requires teaching the sender's AI the features of each smell and stocking matching fragrances in the receiving device.

Possible uses

The researchers envision sharing scents with family and friends who live far apart, adding smells to video viewed through dedicated goggles, and letting online shoppers sample a product's aroma before buying. Perfume is an obvious example: a description such as "fresh" says little about whether you will like the actual scent.

Scentronic is still a prototype, and the list of supported scents is short. But it shows a practical route to digital smell: classify at one end, recreate at the other.