- Researchers programmed robots to make meals according to cooking videos
- Over time you will be able to identify which ingredients work best together
Gordon Ramsay has a new top chef in town in robot form, so you’d better watch out for him.
RoboChef can learn how to make the perfect dish just by watching a cooking video.
Researchers at the University of Cambridge have programmed a machine to make meals that mimics how humans make them.
Using advanced AI, the robot knows every frame what object it’s looking at, whether it’s a vegetable, a hand, or a knife, and how it’s being used.
Over time, you’ll be able to identify which ingredients work best together, and even point out when a human chef might have used the wrong amount.
Robot chefs have been popping up in science fiction for decades, but cooking is a real challenge for robots.
Several commercial companies have built robot chef prototypes, but none are currently commercially available, and they lag far behind human chefs in terms of skill.
“By specifying the materials and how they fit together, we’re trying to create something similar to humans,” said study author and Ph.D. We wanted to know if we could train and learn a robot chef in a step-by-step fashion.” in a plate. It’s amazing how many nuances the robot was able to detect. ”
“Our robot is not interested in food videos that go viral on social media, simply because they are too difficult to follow,” he added.
“But maybe these robot chefs will get better at identifying ingredients in cooking videos faster, allowing them to learn every recipe using sites like YouTube.”
The results of the study, published in the journal IEEE Access, are expected to make robot chefs easier and cheaper to deploy.
The researchers first created a “cookbook” of eight simple salad recipes and filmed themselves making them.
A robot chef trained on a neural network that essentially mimics how the human brain works watched 16 of these videos.
By identifying ingredients and human chef behavior, we were able to accurately determine what was being cooked 93% of the time.
We were also able to detect subtle differences in recipes, such as doubling the portion size.
The robot could also recognize an all-new ninth salad, add it to the cookbook, and make it.
This work was supported in part by Beko plc and the Engineering and Physical Sciences Research Council (EPSRC), part of UK Research and Innovation (UKRI).
