A team of University of Cambridge researchers has developed a robotic chef capable of recreating salad recipes after watching videos of the dishes being prepared by humans.

The team trained the robotic chef using a neural network — which had previously been programmed to identify objects such as fruits and vegetables in the recipes — and by showing the robot several salad recipes.

Source University of Cambridge

Further, computer vision enabled the robotic chef to analyze various objects within each frame of video and to subsequently convert recipes and actions into vectors that served to inform mathematical operations.

According to its developers, the robotic chef, after viewing 16 salad-making videos, accurately recognized recipes 93% of the time. Additionally, the robot could also reportedly detect subtle alterations to recipes.

The robotic chef is described in the article, "Recognition of Human Chef’s Intentions for Incremental Learning of Cookbook by Robotic Salad Chef, "which appears in the journal IEEE Access.

For more information on the robotic chef, watch the accompanying video that appears courtesy of the University of Cambridge.