Gordon Ramsay, Jamie Oliver, and other celebrity chefs might want to pay attention: a cooking robot can learn recipes simply by watching videos.
Researchers at the University of Cambridge programmed a robot to cook like a professional chef.
Using complex artificial intelligence, the robot replicates human actions from videos by processing and remembering every frame. Its main advantage over earlier systems is that it can recognize and pick out the objects it needs to cook — differentiating a vegetable from a hand or a knife, for example, and not confusing them.
The team says the robot will eventually learn which ingredients pair well and could even flag mistakes a human chef makes — like using the wrong amount of an ingredient.
Cooking robots aren’t new. Researchers have long tried to recreate the ideal cooking AI, because cooking is a very complex task for robots.
Some companies have already built prototypes, though none are for sale yet. So far, these machines lag well behind professional chefs in skill.
The author of the study, Grzegorz Sochacki, a research associate at the Cambridge Laboratory of Bio-inspired Robotics, said, “We wanted to see if we could teach a cooking robot to acquire knowledge in the same gradual way humans do — by identifying ingredients and how they combine to create a dish. We were surprised by how many nuances the robot detected.”
He added, “Our robot isn’t interested in the cooking videos that flood social media because they’re too hard to follow. We aim to teach the robot to identify ingredients from videos more quickly and accurately so it can, for example, learn recipes from YouTube.”
The team hopes the results, published in IEEE Access, will make cooking robots easier and cheaper to develop and manufacture.
Initially, the researchers created a “cookbook” with eight simple salad recipes and filmed the cooking process. A cooking robot, equipped with neural networks that mimic the functioning of the human brain, watched 16 such videos. By identifying the ingredients and actions of the human chef, it was able to accurately replicate 93% of the process.
It also identified small variations in the recipes, such as a double portion.
The robot recognized a completely new ninth salad, added it to its cookbook, and prepared it.
The research was partially supported by Beko plc and the Engineering and Physical Sciences Research Council (EPSRC), part of UK Research and Innovation (UKRI).
What other professions will robots take over?
Workers in physical jobs face the greatest risk of being displaced: machine operators and fast-food workers are the most likely to be replaced by robots.
Consulting firm McKinsey analyzed how many jobs could be lost to automation and which professions are most at risk.
The report says machines will perform data collection and processing more efficiently.
That could displace many workers—loan officers, paralegals, accounting specialists, and back-office transaction processors.
The most “human” jobs are those that take place in unpredictable environments.
The report says jobs like gardener, plumber, nanny, and caregiver are less likely to be automated before 2030 because they’re technically difficult to automate and often low-paid, making automation less profitable.
