What if a robot working in a restaurant kitchen could see the way every other cook does? On October 15, Miso Robotics, the leading robotics and technology provider for modern restaurants and hospitality, debuted exactly that.
Before a packed house for LA Tech Week of leading technologists, investors, operators and journalists at its Los Angeles robot lab, Miso unveiled the next generation 3D vision for its Flippy fry station robot. The new vision leverages Miso’s unique collaboration with NVIDIA.
Guests saw where the newest generation of Flippy is today and where it’s headed – side by side. Currently deployed nationwide at White Castle, NBA arenas, California State University campuses, and more, today’s Flippy cooked nonstop at the event, serving fried crowd-pleasers like chicken strips, fries, mozzarella sticks and jalapeño poppers. Beside it, Flippy’s laboratory test unit debuted the new 3D vision system, while a large screen let guests watch in real time what Flippy was seeing.
The moment Flippy dropped food into a fry basket, the robot immediately identified it and knew how to cook it. It even delineated between lookalike foods, like mozzarella sticks and French toast sticks – items that can trip up a busy person at the end of a long shift. When a metal fork was dropped into a fry basket, Flippy spotted it instantly and refused to cook. And Flippy tracked all of the objects in its ecosystem seamlessly.
“Almost everyone who works the fry station in a commercial kitchen burns themselves. Flippy reduces that to zero, while also reducing fry station labor by almost 90% and freeing up employees to focus on the very human things that deliver excellent customer experiences,” said Rich Hull, CEO of Miso Robotics. “Vision is what makes robotics reliable and safe in the real world, and Flippy’s vision is making a massive leap forward. We’re incredibly proud of our Miso team and our collaborators at NVIDIA, and we’re only getting started.”
“Everybody’s talking AI at all the other events, but [Miso is] the only one I know of showing the robot... The fries are perfect, not overcooked, not undercooked, and the fact that it was able to see and detect utensils... this is really a big disruptor.” said Enigma Valdez, a data analyst and attendee at the event.
The new vision upgrade is built on a simple idea: 3D beats 2D. Until now, Flippy relied on lower-resolution infrared cameras that tracked two dimensional visual markers within its environment. The new system replaces them entirely with high resolution color cameras and new AI models, so Flippy sees and immediately recognizes the whole basket, the food inside it, and the fry station around it in three dimensions. It understands not just what is in front of it, but where it is and how it is positioned.
To build these innovations, Miso started with pretrained AI models from NVIDIA’s NGC catalog and fine tuned them with NVIDIA’s TAO Toolkit, using Miso’s own data from Flippy’s years in live kitchens. The system runs in real time on NVIDIA GPU hardware right at the fryer. Flippy has fried more than 5 million baskets and logged hundreds of thousands of hours in live commercial kitchens, and the new vision system builds directly on that real world experience.
“A commercial kitchen is one of the toughest places on earth for a camera: heat, steam, grease, glare and people constantly moving,” said Jeff Faneuff, VP of Software Engineering at Miso Robotics. “We built this system for exactly those conditions. Solving vision here doesn’t just make Flippy smarter today. It speeds up everything we build next.”
The new vision system is designed for kitchens across restaurants and hospitality, from quick service restaurants to stadiums, theme parks, hotels and college campuses. Operators are ready for this kind of progress: according to the National Restaurant Association’s 2026 State of the Industry report, 98% of operators name unsustainable labor costs a top concern.
Miso’s new vision system will be included in upcoming Flippy installations, with first shipments expected in December. To learn more, visit misorobotics.com.