USA | Chef Robotics has advanced its bi-manual physical AI system for prep table food assembly, powered by a food foundation model.
Chef Robotics has announced the development of a bi-manual physical AI system for prep table food assembly. While today’s Chef robots handle high-volume meal assembly on food manufacturing conveyor lines, this new bi-manual physical AI system will focus on lower-volume, higher-complexity prep-table-based assembly for industries such as ghost kitchens, fast-casual restaurants, airline catering, schools, hospitals, military, prisons, stadiums, corporate dining, and hotels.
With the advent of physical AI and imitation learning, Chef’s AI team is developing a new physical AI system designed to handle meal assembly tasks on prep tables, such as back-of-house burger or burrito assembly. These tasks are lower-volume but higher-complexity than food manufacturing on conveyor lines because a single worker (or robot) must assemble the entire meal, rather than breaking the process down into separate workstations for each ingredient.
To perform higher-complexity tasks, the new system will use two robotic arms, enabling bi-manual control. It will be able to perform coordinated, dexterous manipulation comparable to that of human arms and hands. The system’s end effectors will be flexible enough to pick up different food ingredients and utensils.
The new physical AI system will be powered by Chef’s Food Foundation Model (FFM), which learns faster and adapts to a wider range of use cases than traditional robotic systems.
Off-the-shelf vision-language-action models (VLAs) and physical AI models aren’t sufficient for food manipulation. Most VLAs and physical AI models are trained on rigid-body manipulation, but food manipulation involves highly variable, deformable materials (e.g., wet, sticky, irregular items). This requires Chef’s AI models to generalise across a broad range of physical states and interactions.
Instead of requiring separate models for tasks such as picking and placing food, detecting trays, compartments, and inserts, and handling scoopable or discrete ingredients, the FFM supports all of these capabilities through a single “foundational” AI model. It can also be extended to new tasks more efficiently and with improved performance.
Rather than being programmed, the FFM learns from demonstration (imitation learning) to perform specific tasks like assembling a burger or building a burrito bowl. It also generalises across different robotic hardware platforms by learning task representations that transfer across hardware embodiments (such as systems with different kinematics, end effectors, and configurations). In that sense, Chef is building the physical AI layer for food.
The FFM is expected to unlock additional capabilities over time. For example, it may support zero-shot or few-shot ingredient onboarding, adapting to new ingredients with minimal training. The model will also self-improve and autonomously increase yield and consistency over time.
Subscribe to the Restaurant & Café newsletter for weekly industry updates and event coverage.
