In recent years, self-learning robots have been widely used in the field of food processing. Recently, for example, a team of researchers at the University of Cambridge succeeded in giving robot chefs the skills to make omelettes. We’ve seen robots in California that automatically flip burgers, and sausage robots in Berlin. If the technology is popularized, it could greatly free up the cumbersome workload of human chefs.
Video screenshot (from: University of Cambridge)
据悉，该项目由剑桥大学与家用电器公司 Beko 合作开展。科学家们着手将机器人烹饪推向新的领域和高度，因此在探索了披萨、煎饼等应用场景之后，研究团队也在努力思考如何进一步优化机器人的餐点制作技能。
Dr Fumiya Iida, from the Department of Engineering at the University of Cambridge, who led the new study, said: “Cooking is a pretty interesting topic for robotics experts, because human food can never be 100% objective, so the task for scientists is to assess how robots can do better.”
The robotic chefs used in the study have been trained to perform simple prep preparations, from tapping eggshells, stirring, and eventually smearing. Cambridge researchers have equipped it with machine learning algorithms to help it raise the level of omelettes based on subjective feedback.
“Another challenge we face is the subjectivity of human taste, which is that it is difficult for humans to give absolute taste standards,” says Fumiya Iida. With this in mind, robots need to constantly adjust the batch machine learning algorithms to continuously improve and improve diner satisfaction.”
Can robots make omelettes (via)
By combining these assessments with machine learning algorithms, robots can continuously improve the quality of their omelettes over time (quantitative improvements), as the end result confirms. Fumiya Iida says the overall taste is good.
Obviously, as an ordinary but never-before-usual dish, robots need to weigh a combination of color aromas and texture shapes, making them an ideal testing scenario for testing the robot’s machine learning capabilities.
Details of the study have been published in the recent lying IEEE Robotics and Automation Letters.
Originally published as Improving Robotic Cooking Using Batch Bayesian Optimization.