More salt? Robotic chef learns to taste test as it goes
We're starting to see robots gain footholds in the food industry in some pretty interesting ways, from droids that carry out deliveries, to systems that churn out 300 pizzas an hour to cybernetic chefs that single-handedly operate fry stations. Researchers at the University of Cambridge have been tinkering away at the edges of this field of robotics and developed a machine with an ability to "taste test" food as it goes, making sure the balance of flavors is just the way it should be.
The robot chef developed by the scientists is actually a continuation of a project we looked at back in 2020, in which the University of Cambridge team collaborated with domestic appliance company Beko on an interesting concept. The idea was to not just have a machine prepare a pizza or burger, as we've seen before, but have it produce the best meal possible based on human feedback.
Obviously everyone's tastes are different, and to cater to the inherent subjectivity in what makes a tasty meal the researchers developed a new kind of machine learning algorithm. Giving the robot feedback from human samplers enabled it to improve its product over time, tweaking its methods and whipping up an omelette that in the end "tasted great."
Now looking to give the robot its own taste-testing abilities, the scientists have again teamed up with Beko to produce a new and improved version. In doing so, the team sought to mimic the chewing process in humans, which not only physically breaks down food for easier digestion, but floods our mouth with saliva and enzymes that alter its flavors.
Evolved over millions of years, this process also sees the saliva carry chemical compounds from the food to taste receptors on the tongue, which sends signals onward to the brain where it is determined whether something tastes good or not. If a robotic system can do something similar, it could make adjustments to its cooking on the fly, ultimately winding up with a better dish at the end with less human intervention.
“When we taste, the process of chewing also provides continuous feedback to our brains,” said study co-author Dr Arsen Abdulali. “Current methods of electronic testing only take a single snapshot from a homogenized sample, so we wanted to replicate a more realistic process of chewing and tasting in a robotic system, which should result in a tastier end product.”
The team's new machine uses a conductance probe as a salinity sensor, fixed to a robotic arm. The robot was then presented with nine different variations of scrambled eggs and tomatoes, with different amounts of tomatoes and salt in each dish.
The robot was able to "taste" the meal, with the dishes then put through a blender several times to mimic chewing and allow the robot to continue taste-testing it at different stages of the process. The different readings taken by the robot enabled it create taste maps of the dishes in a grid-like fashion, based on the saltiness levels of different "bites."
The scientists hope to add yet more functionality to their robotic chef, planning to work on new sensing abilities that enables it to taste sweet and oily foods.
“When a robot is learning how to cook, like any other cook, it needs indications of how well it did,” said Abdulali. “We want the robots to understand the concept of taste, which will make them better cooks. In our experiment, the robot can ‘see’ the difference in the food as it’s chewed, which improves its ability to taste.”
The research was published in the journal Frontiers in Robotics and AI.
Source: University of Cambridge via EurekAlert