"Very nice, but the last taste was 0.0024% more buttery."

Robots get to have all the fun. In this case what I’m calling a robot is the electronic tongue created by Spanish scientists to taste test the country’s wine. Better than your snobby cousin, this ‘robot sommelier’ raises its electronic pinky in the air and accurately differentiates between three different types of wine.

The wine under study is Cava. Produced in the northeast corner of the country, it’s a sparkling wine that derives its name from the Catalan word for the cave-like cellars where the wine is stored during fermentation. Cava is an ideal wine with which to test the electronic tongue palette because types are categorized according to the amount of sugar added: ‘brut nature’ only has residual sugar while ‘sweet’ has at least 50 gL of sugar. ‘Extra brut,’ ‘brut,’ and ‘extra dry’ have intermediate and increasing amounts of sugar. That a single ingredient can be used to distinguish between the different types makes the task feasible.

The electronic tongue was created by the Group of Sensors and Biosensors at the Universitat Autònoma de Barcelona. Just as with a real tongue, the biologically-inspired electronic tongue is activated by binding to tastant molecules – in this case sugar molecules. It’s an array of voltammetric sensors that sends an electronic signal with a particular pattern that corresponds to the sugar concentration. Like the electrical signals passed on through neurons from the tongue to the brain, the signals generated by the electronic tongue have to be interpreted. For that the group used an artificial neural network that mathematically models biological neural network function. Just like real neuronal networks, the artificial network needs to learn to do its job correctly. The scientists, led by professor Manel del Valle, put the system through repeated training trials so it could correctly associate network patterns with sugar concentrations and type of wine. De Valle explains to the Daily Mail, “It’s a complex training system. You need to show it samples – teach it like you would a child and, once trained, it tells you what a new sample looks like or resembles. Then it can be trained for almost any situation.” At the time they published their paper the electronic tongue successfully discriminated three of the Cava wines. They plan to continue training until they can identify all types of Cava on the market.

So why the need for an electronic tongue? Their intention is not solely to threaten the job security of human sommeliers, but to improve winemaking by detecting defects during the wine elaboration process.

The UAB group aren’t the only ones using robotics to emulate human sensory function. Hossam Haick at the Israel Institute of Technology has created an electronic nose. But rather than assessing a Cava bouquet it acts as a breathalyser for detecting cancer. Because of the risk of deadly leaks aboard a space shuttle the Jet Propulsion Laboratory has developed ENose which detects the concentrations of select chemicals to which it has been trained. Both of these are great examples of bio-inspired design outperforming biology. Even though we don’t know what compounds are specifically produced by tumors, Haick’s electronic nose can still sniff the tumor out. This is due to the fact that the computer-processed chemical signature of a tumor-exposed breath will look different from the signature of a healthy individual’s breath. Human’s are incapable of detecting the tumor-specific odorants (dogs can, however). And JPL’s ENose has a dynamic range that’s far greater than that of a human nose.

Today’s humanoid robots are already equipped with some of the best available visual, auditory and tactile sensory capabilities. When electronic noses and tongues become more generalized no doubt they’ll be added to the robots’ sensory repertoire.

“Taste this. Tell me if it’s bad.”

The applications are endless for an efficient, high-throughput chemical analysis system. I know a lot of researchers who would love to never have to do another labor-intensive and inefficient mass spectrometry. I can also see a use for electronic tongues in materials manufacturing. From the detection of disease, air quality maintenance, and making a finer glass of wine, the young technology can only get better with age.

[image credits: RSC and UAB]
image 1: Nice Bouquet
image 2: Diagram