The Future Of Food Automation: How Chef Robotics Tackles Labor Shortages With Ai Robots

Engineering & simulationdiffusion policyreduced onboarding time from weeks to hours for new tasks

The team uses diffusion policy, a form of imitation learning, to allow engineers to teach the robot new tasks by recording demonstrations rather than writing custom code, significantly reducing the time needed to onboard new ingredient classes.

“diffusion policy and imitation learning is very helpful because now for an engineer, instead of having to write code for a week and then test it, um, or 2 weeks. They can just do a bunch of demonstrations of how to pick up.”

Engineering & simulationimitation learning

The company leverages imitation learning to create an end-to-end model that maps sensor data directly to actuation, allowing the robot to learn from human demonstrations of how to manipulate specific food items.

“we're leveraging imitation learning using a technique called diffusion policy, and that basically allows us to demonstrate to the robot new tasks. And for us, new tasks might be new classes of ingredients.”

Running the company

The company fuses real-world production data collected from deployed robots with imitation learning data to create robust AI models that can handle the physical randomness and variations of ingredients in a manufacturing environment.

“take production data from the field, right? Real production data, and seeing how an ingredients change day by day, and seeing how the randomness in the field affects the robot, take that data, plus the imitation learning data, and kind of fuse it.”