In this regular insideBIGDATA feature we highlight our industry’s movers and shakers, companies that are pushing technology forward, and setting trends for innovation.
We look at companies with a focus on big data, data science, machine learning, AI and deep learning – some new, some old, always leading, always dynamic.
We also take deep dives into new technology promoted (or hyped) as “AI” or my favorite “AI-powered-XYZ” to provide transparency for what’s really going on under the hood.
Watch this column for intimate coverage of some pretty cool firms doing some pretty exciting things.
Enjoy the ride! In this installment of “AI Under the Hood” I introduce “Flippy” by Miso Robotics.
Flippy works in fast-food kitchens, operating a frying station for example.
The product was decades in the making in terms of research in robotics and machine learning.
Flippy is an amalgamation of motors, sensor, chips and processing power that wasn’t possible until just the past few years.
The robotic arm is poised to become a regular fixture in high-volume kitchens nationwide in the coming year.
More economical to employ than a minimum-wage worker, Flippy is designed to fit right in alongside human workers on the fast-food prep line.
Flippy is available for an estimated $2,000 per month subscription.
Early versions of Flippy have been seen in places like Dodger Stadium and at locations of CaliBurger, a small quick-serve chain.
The company has raised more than $13 million in investment capital and is currently seeking another $30 million to gain entry in more fast-food kitchens.
Automation in food preparation could be just the solution for COVID-19 limitations with human staffing.
Miso co-founder Ryan Sinnet, Ph.
has been long entrenched in the robotics space, co-authoring a 2018 research paper “Direct Collocation for Dynamic Behaviors with Nonprehensile Contacts: Application to Flipping Burgers” which appeared in IEEE Robotics.
The research indicates the problem of flipping burgers can be classified as a non-prehensile object manipulation problem.
The paper discusses the Fast Robot Optimization and Simulation Toolkit (FROST), but the software was not used in production for Flippy.
The goal was to obtain joint angle trajectories to achieve two main tasks in the robot: (i) translate burger behavior and (ii) pickup and flip burger behavior.
Translate burger behavior involves moving the Cartesian position of the spatula from one point to another while maintaining friction constraints to prevent the burger from slipping.
Similarly, for pickup and flip burger behavior, the robot has to pick the burger using its spatula and flip in such a way that the under side of burger faces up after the completion of flip.
See the figure below for a representation of the coordinate system involved.
Spatula holding a freshly cooked burger.
Accelerations and orientations along x,y,z directions are shown.
Flippy employs a software platform with a control stack that uses some of the same math implemented in FROST but Flippy also uses a variety of other optimization-based control methods to give the robot a wide range of capabilities.
A funny side effect of the control strategies used is that Flippy’s motions often appear human-like.
“Flippy uses deep learning to locate and identify objects on the grill, as well as equipment and other objects in the kitchen,” said Miso co-founder Ryan Sinnet, Ph.
“We have developed breakthroughs in pose estimation technology using neural networks which allows Flippy to precisely identify the location of objects in the kitchen and pick them up with high reliability.
The theoretical breakthroughs we’ve made were necessary to take low-reliability cutting-edge machine learning research and use it in a commercial environment.
” Sinnet’s research paper described an exploration of a control strategy for picking up and flipping burger patties.
This strategy is something that is part of Miso’s developer tool kit for adding functionality to the system, but this control strategy is not currently used in practice as the company has developed methods with higher reliability for the task of flipping burger patties.
“For Flippy’s real-time scheduling, the objective function can be modified to suit the needs of the restaurant,” continued Sinnet.
“For example, a restaurant might want to ensure customers get their food within 3 minutes.
As another example, a restaurant might want to make sure all food in an order comes out at the same time so half the order isn’t cold.
Flippy is a platform that allows restaurant operators to have freedom over operations by shaping Flippy’s objective function.
” Contributed by Daniel D.
Gutierrez, Managing Editor and Resident Data Scientist for insideBIGDATA.
In addition to being a tech journalist, Daniel also is a consultant in data scientist, author, educator and sits on a number of advisory boards for various start-up companies.
Sign up for the free insideBIGDATA newsletter.