Robot Chef Learns to Mimic Human Tasting Process

Chefs quite frequently rely on taste to assess their dishes while cooking, which might then seem an obvious barrier in the development of robotic “chefs” able to cook truly delicious food. It would not be a simple matter of allowing a sensor to detect, say, salt levels by “tasting” the food once either — human taste receptors receive feedback continuously through the various stages of mastication, creating a complete understanding of the food content as it changes during chewing. Current methods of electronic tasting rely on a single snapshot of a homogenized sample, limiting a robot’s ability to satisfactorily analyze food.

In collaboration with Beko, a manufacturer of domestic appliances, researchers at the University of Cambridge have trained a robot chef to assess the saltiness of a dish through different stages of chewing, approximating the way humans taste. The chef was trained to make omelets based on a human taster’s feedback, creating “taste maps,” which significantly improved the robot’s ability to quickly and accurately assess saltiness compared to other tasting technologies.

Using a mixing method that imitates mastication, the robot “tasted” nine variations of a simple scrambled egg and tomato dish at three different stages of the chewing process — seen in action below. The robot’s arm was equipped with a conductance probe to act as a salinity sensor, which is used to taste the finished dishes and then again after the dishes had been blended to imitate the change in texture caused by chewing. Where other methods are time-consuming and produce only a single reading, this method is cheap and very quickly produces a more accurate “map” of a dish’s seasoning.

Results of the study are reported in Frontiers in Robotics & AI. The technique is currently a proof of concept, but the research team says that through this process which imitates chewing and tasting, robots will be able to produce foods people will enjoy, and will even be able to tweak dishes to individual tastes. Future improvements to the robot chef include the ability to taste different types of food, and sensing capabilities increased to taste for sweetness or oiliness, among other facets of flavor.