Researchers taught a robot chef to taste food during cooking, simulating the process of chewing. The method made it possible to estimate accurately and quickly the amount of salt in the dish, and to make "taste maps", which brought the robot's sensors closer to human perception.
Scientists at the University of Cambridge (UK), in cooperation with Beko, the manufacturer of household appliances, have trained a robot-cook to evaluate the salinity of a dish at different stages of cooking. Previously, the robot had been able to cook an omelette based on taster feedback. Now, it can taste the food itself, simulating the process of chewing, and draw a "map of flavours". The results of the study, published in the journal Frontiers in Robotics & AI, will be relevant to creating methods for automated or semi-automated food preparation. In addition, the new approach has significantly accelerated the assessment of food salinity compared to traditional methods.
Taste perception is a complex process to which food appearance, smell, texture and temperature contribute. Saliva produced during chewing helps to transport chemical compounds to the taste buds that send signals to our brain. Taste changes as we chew, which provides constant feedback to the brain.
Often we taste foods as they are being prepared to judge their taste. Existing electronic tasting methods rely on the analysis of a single homogenised sample. So scientists wanted to reproduce a more realistic process of taste perception in a robotic system.
To do this, the researchers attached a salinity sensor to the robot arm. As it cooked a dish of eggs and tomatoes, it "tasted" the food, taking readings in just a few seconds. To mimic the change in texture that occurs during chewing, the scientists placed pieces of food in a blender and the robot reassessed the taste. The readings collected at different moments in the blender allowed it to create "taste maps" of each dish.
By mimicking human taste perception, robots can learn to cook food that people like and adjust to their individual preferences. Understanding the concept of taste will make robots better cooks. Such devices will be in demand in homes for the elderly, orphanages, hospitals and other institutions, where human resources are not always sufficient to provide all people with a tasty and balanced meal.
In the future, the authors plan to equip the robot with other sensors that will allow it to assess the fat content of food and distinguish between sweet and sour flavours.