Robot 'Chef' Learns To Create Recipes From Watching Food Videos

Researchers have trained a robot to watch and learn from cooking videos and recreate the dish. The robot has been programmed by researchers from Cambridge University. They handed a cookbook to the machine, which had recipes of eight simple salad dishes. After watching a video of a human demonstrating one of the recipes, the robot was able to identify which recipe was being prepared and make it. In fact, after successfully completing the eight dishes, the robot came up with a ninth one on its own, Cambridge University said in a release.

"We wanted to see whether we could train a robot chef to learn in the same incremental way that humans can," researcher Grzegorz Sochacki said in the release. The university also posted a video on its YouTube channel showcasing the culinary skills of the robotic 'chef'.

The robot watched 16 videos, each one showing a human making a recipe. It used computer vision techniques to analyse each frame of the video and was able to identify different objects, such as knife, and other ingredients.

Mr Sochacki said these recipes were not complex and were "essentially chopped fruits and vegetables".

The robot also identified the human demonstrator's arms, hands and face.

Explaining how the robot learned to 'cook', the university release said that the recipes and the videos were converted to vectors and the robot performed mathematical operations on the vectors to determine the similarity between a demonstration and a vector.

"Our robot isn't interested in the sorts of food videos that go viral on social media - they're simply too hard to follow. But as these robot chefs get better and faster at identifying ingredients in food videos, they might be able to use sites like YouTube to learn a whole range of recipes," said Mr Sochacki.

The release said that the robot recognised the correct recipe 93 per cent of the time, and it was able to detect slight variations in each of them, like making a double portion or normal human error.