Listen to this article

Dexai Robotics’ Alfred can handle a variety of food preparation tasks. | Source: Dexai Robotics

Dexai Robotics‘ sous-chef robot, Alfred, is currently prepping salads at the Monarch Dining Facility of Travis Air Force Base in Fairfield, California. The deployment is the first of 10 Alfred robots the Defense Logistics Agency (DLA) will implement as part of the $1.6 million contract.

Alfred uses artificial intelligence (AI) and computer vision to adapt to a variety of surroundings. While the robot is only prepping salads currently at the base, it’s also able to perform a number of other tasks, like scooping ice cream. Alfred is able to switch between different utensils on its own.

“What’s cool about this robot is that it uses utensils, so it can really pretty much prepare anything that the chefs come up with. There are no recipe limitations,” David Johnson, founder and CEO of Dexai, said.

Alfred’s flexibility makes it easy to implement in almost any kitchen without adding additional infrastructure, according to the company. Once installed the robot can get to work the same day.

“I think the thing that takes the longest is working with the chef’s vision,” Johnson said. “So the chef comes up and says ‘Oh, okay, I want to do this. I want to do that.’ And it’s about teaching them what they can do with the robot.”

Dexai robotics was founded in 2018 as a spin-out from the Charles Stark Draper Lab. It’s dedicated to developing market-ready, scalable robotics and AI solutions to address real-world needs. With Alfred, the company takes on a challenge for robotics: handling food.

According to Johnson, food is particularly difficult because of how variable it can be. No two heads of lettuce look the exact same, so a robot can struggle with the uncertainty that comes with prepping food.

“There are difficulties that are specifically related to food,” Anthony Tayoun, founder and COO at Dexai, said. “An example comes from the regulations around food. So, for example, you cannot touch the food directly with automation without fulfilling some requirements.”

“Traditional robotics are designed to do the same thing over and over and over again, without any variation,” Johnson said. “So we’ve built the ability for Alfred to learn from that variation, and then to adapt to it and then be able to serve it.”

Alfred can also help reduce food waste by measuring precise portions, and provide data about exactly how much food needs to be prepped to make a certain number of meals.

According to Johnson and Tayoun, Alfred struggled the most when learning to pick up arugula.

“Alfred learned with time that to pick up arugula effectively it has to push things together so that the greens are all clumped in one corner, and then pick it up,” Tayoun said. “It was just like you would see a young person learning, and that’s just so powerful.”

Dexai hasn’t publicly stated where the other nine Alfred robots will be deployed.