US-based Chef Robotics has achieved National Sanitation Foundation (NSF) certification for its AI-powered meal assembly system, validating its compliance with rigorous food safety and cleanability standards. This means that manufacturers can use the robot to perform “food-safe” handling of ingredients such as frozen peas and corn throughout their production process.

Chef Robotics’ system operates alongside workers on food production lines, picking and placing ingredients from conveyors into trays to enhance food assembly, which accounts for nearly 60-70% of labor in food production.

The robot has been certified against NSF’s commercial food equipment standard. According to NSF, it is the only certified “Food Portioning Robotic Module” within that standard.

“Regulation plays a huge role in food manufacturing, and even though NSF is not a regulatory body but an independent organization, its stamp of approval helps build customer trust,” CEO and founder Rajat Bhageria, tells Food Ingredients First.

“Plus, given it’s recognized globally, it helps alleviate food safety and sanitation concerns for large corporations that operate across different markets.”

Tackling “inconsistent” food textures

The certification applies to Chef Robotics’ C-001748 module. Bhageria tells us the system can handle various food ingredients, such as frozen corn and peas, ravioli, and sauces.

“We recently added curries to our list of ingredients. These are especially difficult because of their saucy consistency (which can cause spilling and splashing) and protein chunks that are often unevenly distributed in the hotel pans our robots pick from.”

Additionally, ingredient inconsistencies such as shape, moisture, or fragility also present operational challenges.

“That’s the bread and butter of our business. Many other food robotics companies have failed because food is hard to manipulate.”

Chef Robotics’ meal assembly system uses computer vision to identify trays on the conveyor in real time, including their position and rotation.

“It then picks at the correct pose (i.e., the position and orientation of the end effector) in the hotel pan that will allow it to take the right amount of food. The robot then uses computer vision again to place the food in the correct location inside a tray,” he continues.

“AI plays a crucial role in all of the decisions our robots need to make, and the key comes back to the real-world data flywheel; an AI model is only as good as the data it’s trained on.”

Beyond simulation

Bhageria explains that traditional robotics and autonomy technology rely heavily on simulation and training AI models on vast amounts of synthetic data before deploying them in the real world. 

But because food ingredients are “wet, sticky, and deformable, physics-based simulation doesn’t work here,” he points out. 

“We quickly realized this when we founded Chef and decided to go all in on real-world data. Every new customer who onboards with Chef deploys our robots in real production environments, which provides us with large amounts of real-world data to train our robots’ AI models.”

The training improves the robots’ performance, allowing the company to onboard more complex food ingredients.

In April, the company secured a US$43 million investment to collect more production data in the food and foodservice sector for better model performance.

Advancing the advent of “food robotics”

Bhageria believes the certification will help boost customer trust in robotic systems and expand their applications beyond meal assembly, such as quality assurance (QA).

“Chef has a placement QA feature that allows food manufacturers to automate quality assurance on assembled meals. But I also see this certification pave the way for other solutions beyond the prepared meals sector.” 

In the future, Chef Robotics aims to enable humans to “do what they do best by accelerating the advent of intelligent machines,” says Bhageria.

“Anything that can free up humans from doing uncomfortable, repetitive, or dangerous work and allows them to do more meaningful work instead — be that prepping, cooking, or assembly — will help us achieve this goal.”

The California-based firm is now planning to expand beyond picking and placing ingredients to solutions that will increase the efficiency of food production lines.

“The reality is that to date, there has not yet been a massive success story in food robotics; if we can be that success story, we can inspire others to build robotics and collectively advance the advent of robotics,” he concludes.