The researchers trained the robot chef to watch and learn cooking videos and recreate the food itself. Researchers from the University of Cambridge programmed their robot chef with a cookbook of eight simple salad recipes. After watching a video of a person showing one of the recipes, the robot was able to identify which recipe was being prepared and cook it.
Also, the video helped the study gradually refresh the cookbook. At the end of the experiment, the robot independently came up with the ninth recipe. The results were published in the journal IEEE Accessshow how video content can be a valuable and rich source of data for automated food production, simplifying the deployment and cost-effectiveness of robot chefs.
Robot chefs have been talked about in science fiction for decades, but in reality, cooking is a difficult task for a robot. Many commercial companies have produced prototype robot chefs, but none are currently commercially available, and they lag far behind their human counterparts in skill.
Human chefs can learn new recipes through observation, whether someone else is cooking or watching YouTube videos, but programming a robot to cook a variety of dishes is expensive and time-consuming.
A human demonstrates one of eight pre-programmed recipes for a robot chef using a neural network. Credit: University of Cambridge
“We wanted to see if we could teach a robot chef to learn step-by-step the way humans do, identify ingredients and learn how they come together on a plate,” said Grzegorz Sohacki of the Cambridge School of Engineering. your paper.
Sochatskyi, Ph.D. The candidate and colleagues in Professor Fumiya Iida’s biotechnology-inspired robotics lab developed eight simple salad recipes and filmed their making. They then used a public neural network to train their robot chef. The neural network was programmed to identify a number of different objects, including fruits and vegetables used in eight salad recipes (broccoli, carrots, apples, bananas and oranges).
Using computer vision techniques, the robot analyzed every frame of the video and was able to identify various objects and features such as knives and materials, as well as the hands, arms, and face of the human demonstrator. Both the recipes and videos were converted to vectors, and the robot did mathematical operations on the vectors to determine the similarity between the demonstration and the vector.
An example of a robot that produces “salad” by watching a person prepare it. Credit: University of Cambridge
The robot recognized the correct recipe in 93% of the 16 videos it watched, but only detected 83% of the human chef’s actions. The robot was also able to detect that minor changes to a recipe, such as making a double batch or making a simple human error, were variations rather than a new recipe. The robot also correctly recognized the introduction of the new, ninth salad, added it to its cookbook, and prepared it.
“It’s incredible that the robot can pick up so many nuances,” Sochatsky said. “These recipes aren’t complicated – they’re basically just sliced fruit and vegetables – but it was very effective in realizing, for example, that two sliced apples and two sliced carrots are the same recipe as three sliced apples and three chopped carrots.”
The videos used to train the robot chef are not videos like cooking videos made by some social media phenomena, full of fast cuts and visual effects, that move quickly between the person preparing the food and the food they are preparing. . For example, when the human pointer wraps his arm around the carrot, it will be difficult for the robot to recognize a carrot. In order for the robot to recognize the carrot, the demonstrator had to hold the carrot up so the robot could see the whole vegetable.
“Our robot isn’t interested in cooking videos that go viral on social media – it’s very difficult to watch them,” Sochatzki said. “But as these robot chefs get better and faster at identifying ingredients in cooking videos, they can use sites like YouTube to learn a wide variety of recipes.”
Source: Port Altele
As an experienced journalist and author, Mary has been reporting on the latest news and trends for over 5 years. With a passion for uncovering the stories behind the headlines, Mary has earned a reputation as a trusted voice in the world of journalism. Her writing style is insightful, engaging and thought-provoking, as she takes a deep dive into the most pressing issues of our time.