Watch a robot make an omelette: University of Cambridge and Beko team up to research kitchen of the future
How do you like your eggs? Chances are, you’ve not had them cooked like this before.
Engineers at the University of Cambridge have trained a robot to prepare an omelette - from cracking the egg to plating the finished dish - to help Beko prepare for the ‘kitchen of the future’.
The researchers collaborated with the domestic appliance company to show how machine learning could be used to train a robot not just to prepare a dish, but to take account of the subjective matter of taste.
Dr Fumiya Iida, from Cambridge’s Department of Engineering, who led the research, said: “Cooking is a really interesting problem for roboticists, as humans can never be totally objective when it comes to food, so how do we as scientists assess whether the robot has done a good job?”
From the manipulation of a variety of tools to the computer vision required to ‘see’ the objects being manipulated, cooking represents a significant challenge for robots.
They need sensing capabilities to ensure they pick up ingredients and objects with appropriate force and need to be able to cope with human interaction.
And they must achieve this while aiming to produce something that eludes many of their human counterparts - a consistent end product.
Little wonder then, that despite long-held dreams of robotic kitchens, none is yet commercially available, although one company, Moley Robotics, has pledged to change that this year.
While other research groups have trained robotic chefs to make cookies, pancakes and pizza, none has yet optimised them for the subjective nature of taste.
To explore this, the Cambridge researchers picked the omelette for their test dish.
“An omelette is one of those dishes that is easy to make, but difficult to make well,” said Dr Iida. “We thought it would be an ideal test to improve the abilities of a robot chef, and optimise for taste, texture, smell and appearance.”
Working at the Department of Engineering, using a test kitchen supplied by Beko plc and Symphony Group, Dr Iida and his colleagues trained their robot chef to crack open an egg over a bowl, tip in salt and pepper, then whisk, before tipping it into a frying pan and carefully stirring over heat, until it was ready to plate up.
They deployed a machine learning technique that utilises a statistical tool known as Bayesian Inference to derive information from the limited data feedback achieved from human tasters who, after all, could only eat a limited number of omelettes.
“Another challenge we faced was the subjectivity of human sense of taste - humans aren’t very good at giving absolute measures, and usually give relative ones when it comes to taste,” said Dr Iida. “So we needed to tweak the machine learning algorithm - the so-called batch algorithm - so that human tasters could give information based on comparative evaluations, rather than sequential ones.
The study demonstrated that machine learning can be used to obtain quantifiable improvements in food optimisation and such an approach could easily be extended to multiple robotic chefs.
“The omelettes, in general, tasted great – much better than expected!” said Dr Iida.
Further studies are needed to investigate other optimisation techniques and their viability.
Dr Graham Anderson, the industrial project supervisor from Beko’s Cambridge R&D Centre, said: “Beko is passionate about designing the kitchen of the future and believes robotics applications such as this will play a crucial part. We are very happy to be collaborating with Dr Iida on this important topic.”
The results, reported in the journal IEEE Robotics and Automation Letters, will be available as part of the virtual IEEE International Conference on Robotics and Automation (ICRA 2020).