Called Scentronic, the system combines AI with gas sensors and a small device that releases controlled amounts of scented liquid.
The system uses 6 sensors to detect and classify odors. AI algorithms then identify the source from 5 tested categories: beer, cake, coffee, orange and red wine. The results are transmitted over the internet to a device that releases the corresponding scent.
The researchers developed several designs, including wristbands, necklaces and handheld devices. They also built a cup that combines smell and taste stimulation. In a demonstration, selecting virtual beer or coffee in a mixed reality environment triggered the cup to release the corresponding aroma and taste cues.
Tests involving 14 participants showed 72.86% accuracy in identifying five odors. Participants distinguished combinations of two scents with 71.43% accuracy. Their accuracy in ranking 5 levels of alcohol scent intensity was 58.57%.
The technology could eventually support remote smell tests, allowing clinicians to assess patients’ ability to detect odors and monitor changes in their sense of smell. It could also help make virtual reality, online shopping and digital communication more immersive.
However, Scentronic cannot yet reproduce smells with complete accuracy. It identifies an odor category and recreates it using prepared scent options rather than copying the exact chemical makeup of the original smell. The researchers say the system needs more sensors and scent channels to recognize and reproduce a wider range of odors.
Earlier, Qazinform News Agency reported that smelling coffee might ease anxiety and fatigue.