IBM’s e-tongue can taste and identify liquids, but can it talk? (Yes, sorta!)

Hypertaste, IBM's sensor for fingerprinting liquids, could be used to ID counterfeit wines. (IBM Research)

IBM Research is developing Hypertaste, an AI-assisted portable electronic tongue that can taste and identify liquids quickly and at lower cost than conventional lab tests.

The chemical sensor could be used for a water quality check of a lake or stream in the wilderness, or by a food maker wanting to identify a counterfeit wine, according to a blog.

Healthcare providers could also use Hypertaste to fingerprint liquids such as a person’s urine, which constantly changes depending on lifestyle and nutrition, making a series of snapshots with the sensor valuable. “Such a tool could allow sub-grouping of patients in chemical trials for new drugs by matching the individual responses of patients to a treatment with information on their personal metabolomes,” wrote Patrick Ruch, research staff member, in the blog. “The spectrum of possible applications is vast.”

An IBM video depicts Hypertaste as a small circular device that looks like a slice of lemon that can be hooked on the edge of a beaker to identify a liquid. Inside of Hypertaste are off-the-shelf electronics and swappable polymer-coated electrochemical sensors. The electronics measure voltages across electrodes in an array. A liquid sample can be identified in less than a minute with results transmitted wirelessly to a smartphone or other portable device.

IBM gets that speed by using cross-sensitive sensors with intelligent software that can be outsourced in the cloud. Using the cloud means that the sensors can be reconfigured from anywhere without altering the hardware. Sensors will also be able to learn from one another by exchanging information about new liquids they encounter in an AI and machine learning framework, IBM said.

It will be hard to fool Hypertaste because it doesn’t rely on identifying any single substance in a liquid. Instead, it uses a “combinatorial” sensing model, which resembles human natural senses of taste and smell where people don’t have a receptor for each molecule in food or drink. Instead, "combinatorial sensing relies on individual sensors to respond simultaneously to different chemicals,” the blog said. “By building an array of such cross-sensitive sensors, one can obtain a holistic signal, or fingerprint, of the liquid in question.” 

The electrochemical sensors in Hypertaste are comprised of pairs of electrodes and they respond to the presence of a combination of molecules with a voltage signal. The combined voltage signals of all the pairs of electrodes represent the liquid’s fingerprint, which is compared to known chemicals in a database in the cloud for identification.

“We are confident that, through forthcoming refinements, the use of AI-assisted, portable chemical sensors will meet the needs of many industries when it comes to fast and mobile fingerprinting of complex liquids,” the blog concludes.

In that sense, the Hypertaste e-tongue both tastes and talks.

RELATED: “Sensor provides cleaner way to detect urinary tract infection”

Suggested Articles

UBS says Intel may outsource 10nm wafer production

Applications for Toposens’ “Bat Vision” TS3 sensors include navigation systems that mimic the echolocation techniques employed by bats and dolphins.

Designers can initiate projects before licensing with new Arm Flexible Access program