We write frequently about assistive technologies for impaired vision, hearing, touch, and balance. Fortunately, it’s relatively easy to create word pictures to explain the basics of senses that, while not strictly binary, can be quantified. Sure, there are nuances of sound, touch, and even balance and vision, but on the crudest levels, you either see, hear, or feel something or you don’t. You “have” your balance, or you don’t. Taste and smell involve many compounds singly and in combinations. The sense of taste can be quantified and reported as so much of this and so much of that, but the number of variables significantly increases the challenge in designing human-made taste or smell sensors. IBM Watson has solved this for taste.
IBM Research-Zurich created Hypertaste, an AI-powered electronic tongue that identifies tastes much the way humans do, by recognizing chemical combinations. Hypertaste fills the gap between small, inexpensive, portable technologies that measure single properties such as pH sensors and huge, expensive, stationary machines like nuclear magnetic resonance (NMR) spectroscopes. The Hypertaste sensors don’t measure individual elements or compounds. The Zurich researchers synthesize polymer coatings for each of Hypertaste’s electrochemical sensors. The sensors comprise pairs of electrodes that measure the voltage of combinations of molecules.
According to IBM research staff member Patrick Ruch, the Swiss team uses off-the-shelf electronics to measure the voltages from arrays of electrochemical sensors. The device transmits the voltage data to a mobile device which relay’s the information to a cloud server. In the cloud, a machine learning classification algorithm compares the data signature with those of known liquids and sends that information back to the mobile device. The Hypertaste app displays the “taste-test” result in less than a minute, Ruch says.
Potential applications for Hypertaste include industrial testing and inspection, food and beverage testing and development, environmental monitoring, and pharmaceutical testing. Hypertaste could also be used to assess a patient’s metabolomic fingerprint, which represents the body’s total molecular content.
IBM Research-Zurich presented a descriptive study of Hypertaste at the 2019 ISOCS/IEEE International Symposium on Olfaction and Electronic Nose (ISOEN).
