At Creator in San Francisco, California, USA, a glass-faced machine funnels brioche buns down a ramp, where they're sliced, toasted and topped. Another part of the machine grinds and cooks hamburger patties, before slapping them on top. Human hands are nowhere to be found in the cooking process. Instead, the machine is fully automated, the result of an eight-year project to develop a robot capable of handling every human hamburger-making task—faster and more efficiently.
“What you're watching is a technological feat,” Alex Vardakostas, Creator's CEO, told Bloomberg. That technology, which was on display for the restaurant's September opening, isn't perfect. But that's not the point, he said. “Self-driving cars need a lot of mileage before they become reliable. You're seeing something similar here.”
Project leaders around the world seem to agree, as they launch ambitious plans to build robots that can automate much of the restaurant kitchen. British firm Moley Robotics is developing a robot capable of learning new recipes. In China, a nearly four-year, CNY10 million project to develop robots that cook authentic Hunan cuisine grabbed national headlines. And in the United States, Creator has plenty of competition, from Spyce in Boston, Massachusetts to Zume Pizza in San Francisco, California.
There's good reason the food industry is ripe for robot-development projects. Like farming and manufacturing—other industries reshaped by robotics—food prep involves repeatedly performing the same tasks, making it a promising niche for automation. According to an analysis by McKinsey & Company, 73 percent of the activities food service and accommodations workers now perform could be automated.
Source: McKinsey & Company
“Simple tasks accumulate and put restaurant and hospitality workers in a stamping-out-fires kind of mindset. Embedding robotics to do these simple tasks can improve how their daily work is performed,” says Juan Higueros, co-founder and COO of Bear Robotics, San Mateo, California, USA.
—Juan Higueros, Bear Robotics, San Mateo, California, USA
But scaling the new technology—so that it's fast, fail-proof and flexible enough to respond to customer requests—is easier said than done. To really deliver value, Mr. Higueros says, these projects must push past novelty and consider how their initiatives will be implemented across hundreds (or thousands) of locations.
From Snack to Scale
Flippy, developed by Miso Robotics in Pasadena, California, is one early product making the leap from novelty prototype to wide-scale implementation. The industrial robotic arm uses cameras, artificial intelligence, thermal scanners and lasers to know when to turn burgers and remove them from the grill, complete deep-frying cycles or chop vegetables. But, when Flippy was first installed in one of CaliBurger's kitchens in March, problems quickly emerged.
“Mostly it's the timing,” Anthony Lomelino, CTO of Cali Group, which operates the CaliBurger chain, told USA Today. “When you're in the back, working with people, you talk to each other.” The robot instead worked at its own cadence and would consistently place burgers on the wrong tray. Rather than proceed with the rollout to additional CaliBurger restaurants, the project team retooled the robot's pacing and designed a training program that would help kitchen workers understand how to better work alongside Flippy.
Two months later, the revamped robot was back in the kitchen—along with a program to retrain restaurant staff “to serve as ‘Chef Techs’ that work alongside Flippy and monitor the related software and hardware system,” Mr. Lomelino told Hospitality Tech. CaliBurger plans to have Flippy installed in the majority of its U.S. locations by the end of 2019.
Training kitchen staff and tweaking robots to work together is but one challenge of widespread rollouts. On the technological side, the task of building robots beyond singular duties is still a monumental one, says Henny Admoni, PhD, assistant professor at Carnegie Mellon's Robotics Institute, Pittsburgh, Pennsylvania, USA. Cooking in a restaurant kitchen involves thousands of small movements—like opening a refrigerator and grabbing a jar—that humans take for granted. “People have a tremendous capability to recognize and manipulate objects,” she says. “Robots are still struggling to do those kinds of tasks.”
In April, Carnegie Mellon University partnered with Sony researchers on a four-year project to develop an entirely robotic counter-to-table dining experience. To build robots capable of more than unitasking, project leaders had to first bring together the right cross-functional team, including experts in artificial intelligence, machine learning and human-robot interaction.
Dr. Admoni doesn't imagine robots will ever fully replace kitchen staff in every restaurant. As Panera CEO Blaine Hurst told The Wall Street Journal, “There's a craft to making food, and that's hard to replicate with robotics.” —Ashley Bishel
Bots Beyond the Kitchen
Back-of-house isn't the only hot spot for bot projects. In July, Alibaba opened a seafood restaurant, Robot.He, with an entirely robotic waitstaff. Chicago-based chain Wow Bao opened an automated eatery in December 2017, allowing diners to pay via app or kiosk and then collect their cooked dishes from LED-lit cubbies. And Domino's and Ford have a pilot project underway in Miami, Florida, USA, using self-driving cars to deliver pizza. Data from the pilot project could drive a larger initiative to launch a fleet of self-driving delivery vehicles across the country.
Robots can help drive out inefficiencies across the dining experience, says Juan Higueros, cofounder and COO, Bear Robotics, San Mateo, California, USA. When his business partner, John Ha, was operating his own restaurant, Mr. Ha couldn't help but notice all of the small, repetitive tasks that might be better executed by a robot—freeing the waitstaff to spend more time interacting with guests and elevating the dining experience.
So, in May 2017, the two launched a project to build that bot. The indoor food runner robot, nicknamed Penny, serves as a bridge between the kitchen and dining room, running food to guests and returning dirty dishes. As part of the prototype's pilot, the project team catalogued 25,000 customer interactions, detailing everything from how much weight the robot can bear to how guests interact with it.
They also surveyed waitstaff and diners for what they'd like to see in a restaurant bot.
“One interesting request we received was that people want Penny to be able to greet guests or say ‘excuse me’ in crowded, narrow spaces,” he says. So the project team incorporated that stakeholder feedback, helping the robot seem just a bit more human.