Close Ad

The Spoon

Daily news and analysis about the food tech revolution

Behind the Bot

Now That They’re in Grocery Stores, Has the Robot Backlash Begun?

Both Walmart and Ahold Delhaize expanded their use of robots this year. However, according to two big news stories in less than a week, their entry into the workforce is off to a rocky start. Is the grocery industry in for a robo-backlash, and what that will mean for the automation in that sector?

In theory, robots are supposed to take over the manual and repetitive tasks, like taking inventory, scrubbing floors or spotting spills and messes. This, in turn, frees up humans for higher-level tasks and the time to engage in more customer service. To this end, in January, Ahold Delhaize ordered 500 Marty robots for its GIANT/MARTIN’S and Stop & Shop stores, and in April, Walmart expanded its use of in-store robots to 1,500 locations. But at least initially, the theory of robots being efficient helpers is running into some harsh realities.

Last Thursday, The Washington Post ran the story “As Walmart turns to robots, it’s the human workers who feel like machines.” In it, The Post chronicled some of the issues the bots have been having in stores including breaking down, functioning erratically, freaking out shoppers and frustrating employees. As The Post writes, it seems like robots are creating the very problem they are supposed to be fixing:

But the rise of the machines has had an unexpected side effect: Their jobs, some workers said, have never felt more robotic. By incentivizing hyper-efficiency, the machines have deprived the employees of tasks they used to find enjoyable. Some also feel like their most important assignment now is to train and babysit their often inscrutable robot colleagues.

Then today, less than a week after The Post story, comes a story from The New Food Economy titled “Stop & Shop now has big, goofy-looking robots patrolling its aisles. What, exactly, is the goal?” It too, talked of its “Marty” robots malfunctioning, having limited functionality to begin with (something we noted at the time of the announcement), and creeping out customers (perhaps because Stop & Shop gave the robots giant googly eyes).

Some of these potential missteps in implementing robots could be because we are still in version 1.0 of this automation experiment, and there seems to be a mismatch between the customer expectations, robot design, and the tasks being handed over to robots.

The robots being used in the front of store (fulfillment robots in the back of house are a different story) are industrial looking. They are tall, cold and utilitarian in design, and move about in a very, well, robotic manner. Shoppers aren’t used to sharing aisles with an indifferent machine that is beaming a light to scan shelves for missing inventory or just watching the floor (with giant googly eyes) to see if anyone has made a mess. The Marty robot, we should note, doesn’t clean up any mess; it just stops and points them out for humans to deal with.

One has to wonder just how long-term these problems with robots will be. There’s a raft of startups looking to retrofit stores with banks of high-tech cameras in the ceiling to facilitate cashierless checkout. These cameras, aided by computer vision and machine learning, could easily take over spotting empty shelves and alerting staff about spills without the need for a robot roving the store. Walmart already debuted this type of invisible inventory management at its IRL store in NY in April.

This eye in the sky approach alleviates any creepy factor associated with bumping into a giant robot as you pick out a pint of Ben & Jerry’s, but it also has you shopping in a surveillance supermarket. As for employee issues, this is why Albertsons brings labor unions to the table at the very start of any automation discussion.

In addition to this being a story about busted robots, this is also a story about change. As the saying goes, the only thing people hate more than change is things staying the same. The way we shop for groceries is undergoing and will continue to undergo big changes in the coming year. Robots will be a part of that, and there will be problems that arise. It will be up to the retailers to figure out the right balance to avoid a robot-driven backlash.

Build Your Own Home Robot Bartender for Under $300 with MyBar.io

We’re in the midst of planning a summer party at our house (the theme is yacht rock), and while I love throwing parties, I’m not a fan of the mess left after an evening of making cocktails. Between people sloppily pouring booze and mixers, it’s a sticky pain to clean up.

Which is why I’m tempted by Mybar.io DIY robot bartender. It’s basically a box that houses up to 9 pumps and tubes as well as some circuitry. You can order one for $299 if you build it yourself, or $399 if you order it fully assembled.

If you buy the kit, there’s an online guide to walk you through how to build it. Once set up, you download the app (Android only, because those tablets are cheaper), place the tubes in your selected bottles, and assign them to a pump via the app (e.g. vodka is pump one, orange juice is pump 2, etc.). Once you have all of your pumps labeled, the app takes stock of what booze and mixers you have and automatically generates a list of drinks you can make. Pick a drink and tap the button and the DIY Bar dispenses your cocktail.

I spoke with Mybar.io Founder Juan Pablo Risso, who told me that he is decidedly going after the DIY market right now. The company debuted the robot bartender at the Maker Faire last month in Portland, OR, and in keeping with that DIY spirit, everything about DIY Bar is open source. You can download the plans for the hardware and the circuit board, as well as the software and firmware from the company’s Github repository.

Right now, Mybar is a side gig for Risso, whose day job is working on IoT for Samsung. Mybar is bootstrapped and Risso said that he’s sold about 50 units since so far.

Risso and Co. are facing some stiff (drinks) competition in the home bartending appliance space. Bartesian (manufactured by Hamilton-Beach) is a Keurig-like countertop cocktail device that also sells for $299 (no assembly required). And on the higher end, Barsys offers a sleek drink making machine, but it will set you back more than $1,000. Which particular cocktail robot you want to buy probably depends on how much work you want to put into your machine, and how much flexibility you want in your mixology.

I just don’t want to spend all my time at my next party making drinks. Perhaps Mybar’s DIY robot is in my future.

Video: Sony’s Masahiro Fujita on Bringing AI and Robotics to Food

We were thrilled when Masahiro Fujita, Chief Research Engineer of the AI Collaboration Office of Sony agreed to be a speaker at our ArticulATE food robotics conference in April. He started Sony’s Robot Entertainment project in 1993 and led the development of the famous AIBO robotic dog.

During his ArticulATE presentation, Fujita talked about how historically, Sony has provided technology throughout the entertainment stack. Its studio Sony Pictures finances movies, filmmakers use Sony products to shoot and create movies, and consumers can watch movies on their Sony Blu-Ray or PlayStation Network.

Fujita said that Sony views the food world in much the same way. It wants to provide the underlying robotic and AI technology that can help creative types like chefs make their food, as well as the mechanisms for people at home to enjoy high-level cooking.

Because the technology is still so nascent and not ready for prime time with consumers, Sony is looking first at B2B applications. The company really wants its AI and robots to be able to make meals in a 3 Michelin star restaurant. Those of us waiting for robotic help at dinnertime at home are going to have to wait awhile.

Check out Fujita’s full presentation (and a glimpse into the robo-chef future) from ArticulATE below:

ArticulATE 2019: Where Is It All Going? A Look Forward with Sony's Masahiro Fujita

Welp. Robots Have Knives Now, and Know How to Use Them (to Slice Onions)

Well, fellow humans, we had a good run, but our time is over. Robots have their knives out — literally — and know how to use them.

Terminator-esque teasing aside, IEEE Spectrum has a video roundup of some of cutting-edge (sorry) robotics research being done right now. Included among the videos is “Robotic Cutting: Mechanics and Control of Knife Motion,” by Xiaoqian Mu, Yuechuan Xue, and Yan-Bin Jia from Iowa State University, in Ames, Iowa, USA.

You may think that having a robot to slice an onion mainly entails a big mechanical arm slamming a knife down, but you’d be wrong. The researchers created a program that combines and coordinates pressing, pushing and slicing motions. From the research paper’s Introduction:

Cutting skills such as chop, slice, and dice are mostly beyond the reach of today’s robots. Technical challenges come not just from manipulation of soft and irregularly-shaped objects, but more from doing so while fracture is happening. The latter requires planning and force control based on reliable modeling of an object’s deformation and fracture as it is being cut. The knife’s movement needs to be adjusted to progress in terms of material fracture. Its trajectory may need to be replanned in the case of an unforeseeable situation (e.g., appearance of a bone).

Robotic Cutting: Mechanics and Control of Knife Motion

As you can see from the video, this particular robot won’t be wowing crowds at a Benihana anytime soon, but it shows once again that robots are getting more proficient at higher-skilled tasks. Automation is coming for food sector jobs, and while we think of them right now in terms of flipping burgers and bussing tables, robots will be automating more and more tasks in restaurants, like prepping vegetables.

Dishcraft, for example, is still pretty tight lipped around what it’s working on, but the company has talked about building robots to do specific tasks in restaurant kitchens like prep work. Miso Robotics’ Flippy was created in part to take over dangerous tasks like working the grill and deep fryer in the kitchen, and the company has already talked about Flippy eventually chopping vegetables.

While there are still many issues to work through with the rise of robots, having them handle knives in the kitchen (and saving countless fingertips from lacerations) is probably not such a bad thing.

Domino’s Has a New AI Tool to Assess Pizza Quality

While it would never classify as an actual problem, there’s something rather annoying about getting a pizza that’s less than a perfect circle or comes with unevenly distributed toppings and sauce. Domino’s, who more and more wants to be a tech company that just happens to sell pizza, announced this week it has begun to address this quality control issue by using AI.

In Australia and New Zealand, the company is debuting its DOM Pizza Checker, which is a smart scanning device that hangs above the cut bench at Domino’s locations and uses AI, machine learning, and sensor tech to assess the quality of the pizzas before they’re sliced and boxed up.

When the pizza arrives at the cut bench, DOM compares its quality to existing pizza images stored in its database and grades the pie based on whether it’s the right kind (e.g., thin crust versus thick) with the right toppings which are evenly distributed. If the pizza doesn’t pass muster, DOM will notify the team of human workers, who will make the pie again.

According to the DOM website, Domino’s has been developing the tech in partnership with Dragontail Systems for the past two years. It joins a growing number of tech offerings the pizza company has unveiled over the last few years, from chatbots taking orders to location technology to in-car ordering.

DOM Pizza Checker is another example of why Domino’s landed on our most recent Food Tech 25 list. The company seems to be endlessly dabbling in new ways to deliver pizzas to your doorstep faster. Some of them, like the AnyWare program, work great. Others are a little less impactful (hello, DXP), and I question how frequently the company slaps the “AI” label on projects. But whether something like DOM is a huge success and heads Stateside or drops into oblivion, its very existence shows that Domino’s is analyzing every angle of the process to try and get a better pizza into your hands.

Ford Developing Bi-Pedal Robot to Carry Deliveries from Driverless Cars to Your Door

Plenty of companies are bringing robot-powered delivery of food and other household goods to the last mile, but most stop at the last few feet. Autonomous cars park at a curb and little rover bots typically can’t climb the front steps of a house.

Which is why Ford is working on a bi-pedal robot that literally walks deliveries from driverless cars right up to your front door(h/t to Bloomberg). Dr. Ken Washington, Vice President, Ford Research and Advanced Engineering, and Chief Technology Officer published a post on Medium today outlining the program, writing:

Enter Digit, a two-legged robot designed and built by Agility Robotics to not only approximate the look of a human, but to walk like one, too. Built out of lightweight material and capable of lifting packages that weigh up to 40 pounds, Digit can go up and down stairs, walk naturally through uneven terrain, and even react to things like being bumped without losing its balance and falling over.

Like something straight out of an Asimov novel, Digit folds up and sits in the back of a driverless delivery van. When a package needs to be delivered, it emerges from the vehicle, stands up and carries the package to a person’s doorstop. Digit doesn’t have a ton of autonomy gear and processing power on it. Instead, the driverless car, which is packed with sensors and mapping equipment, sees the surrounding area and sends Digit the best path to the door. If Digit needs help, or encounters something unexpected, the problem can be sent up to the cloud where another system (perhaps even a human) can assist.

Though the Ford post didn’t mention groceries specifically, they are a good use case for this type of robot delivery. Groceries are heavy, and even if a driverless car brings them to a house, a person still needs to go out to the street to retrieve and lug them back inside. The weight of groceries is one of the reasons self-driving delivery company, AutoX moved more into (the much lighter) restaurant food. For most, this walking to the curb is a minor inconvenience, but for those who have trouble moving, a robot walking packages to the door would be a big help.

Ford isn’t alone in getting goods up directly to your front door. Earlier this year, FedEx unveiled a its own delivery robot that can climb stairs (though it uses wheels, not legs), and Amazon received a patent for an autonomous robot that would live in a home’s garage and would venture out to fetch packages from delivery trucks.

There’s no word on when or if this particular version of Ford’s delivery vision will be coming to a neighborhood near you, there are still a lot of regulatory hurdles for self-driving vehicles to get through. But with the pace of innovation, robots are bound to be bounding up your walkway to deliver a package someday soon.

Ocado Leads $9M Seed Round in Food Robotics Company, Karakuri

U.K. based online grocer Ocado announced today that it has acquired a minority stake in London-based food robotics company, Karakuri. Ocado’s investment led a $9.1 million seed round in Karakuri, which also included Hoxton Ventures, firstminute Capital and Taylor Brothers.

Karakuri makes two different food robots: The DK-One, a more industrial robot that can assemble (not cook) 48 ingredients into ready-to-go meals on a mass scale in commercial kitchens; and the Marley, which is a smaller scale machine meant for applications like candy stores and frozen yogurt dispensing and topping.

Ocado is no stranger to robots: the company uses them to power its smart, automated warehouses, where totes on rails bundle up grocery orders for delivery. With the minority stake in Karakuri, Ocado appears to be setting itself up to expand this robot-powered automation into other forms of food delivery. From Ocado’s press announcement:

The [DK-One] can be used in the assembly of all boxed meals, using a configurable, modular design which can easily be installed in-store or in “dark kitchens”, and can aggregate up to 48 food items to create a wide range of food-to-go options.

Dark kitchens (restaurants that are delivery only) in particular are an interesting avenue for Ocado/Karakuri. Not only could a dark kitchen automate order assembly quickly, but the restaurant could then subscribe to Ocado’s logistics and delivery service to manage and optimize getting those orders to customers. This would mean more revenue for Ocado and also more data, giving the company insight into what, when and where people are ordering different restaurant meals.

Ocado also said it would tie Karakuri’s robots into its existing grocery service, which makes me wonder they will be used for something akin to customized meal kits, or even prepared food that customers could shop for as part of their daily or weekly shopping.

As we saw at our ArticulATE conference last month, automation is invading almost every part of the food stack. Here in the U.S. companies like Takeoff, Alert Innovation and Common Sense Robotics are creating robot-powered micro-fulfillment centers for grocery stores to speed up online order processing. Kroger, which is an investor in Ocado, is building out Ocado-powered smart fulfillment centers here in the U.S. to speed up its own grocery fulfillment and delivery. Will that now include Karakuri robots?

Ocado said that it would take delivery of its first Karakuri robot in the second half of this year. For its part, Karakuri said it will use the new money to further develop its technology, “strengthen its IP base,” and expand its team.

Invert Robotics Raises $8.8M for its Wall Crawling Inspection Robot

Robots are best for jobs that are any combination of the three Ds: dull, dirty or dangerous. Inspecting industrial food and beverage tanks seems like all of those wrapped into one, which helps explain why Invert Robotics just closed an $8.8 million funding round, the company announced today.

Based in Christchurch, New Zealand, Invert Robotics builds wall-crawling robots that can inspect the insides of confined or difficult-to-reach spaces like storage tanks at food and beverage processing facilities. The robot itself looks like a Spinal Tap Stonehenge-sized military tank and uses proprietary suction cups with a friction-reducing coating to climb up just about any wall, including non-magnetic ones.

Once inside, the operator can use the on-board camera can inspect the inside of the tanks for cracks, bad seals or improper cleaning. These types of inspections are important because cracks in tanks can capture bacteria, thus contaminating anything held inside a tank. Additionally, spotting such defects inside tanks early can help prevent industrial accidents like tank failures that result in spills.

By using wall-crawling robots, Invert says it can perform inspections better and more freely than a human can. First, tank inspections typically require a scaffolding be built or rope harnesses used by a person inside the tank. Invert’s robot just sticks to the wall of the tank, and because it is so small, it can navigate into smaller spaces than a human would typically fit. Additionally, the robot is more hygeinic, as a person rummaging around inside a tank could introduce some kind of bacteria or pathogen that could contaminate future contents. Invert says all this translates into faster inspections and less downtime for the tanks, which is important for a processing facility.

This funding round was lead by Finestere Ventures, and brings the total amount raised by invert to $15.9 million. Invert currently has more than 50 customers worldwide and this new money will go towards opening up a U.S. office and hiring up from 40 to 65 employees. The new cash will also go towards research and development to turn Invert’s robot into more of a platform with added functionality like ultrasound to detect wall thickness and Eddy currents for surface and sub-surface inspection.

With today’s announcement, that little wall crawling robot just brought in another d: dollars.

Washington State to Allow Delivery Robots on Sidewalks

If you live in the state of Washington, you could soon be walking alongside delivery robots on the sidewalk, thanks to new legislation signed today by Washington governor Jay Inslee.

House Bill 1325 creates a regulatory framework that personal delivery devices (a.k.a. delivery robots) must follow. The new legislation defines personal delivery devices as:

  • Intended primarily to transport property on sidewalks and crosswalks
  • Weighing less than one hundred twenty pounds
  • Operating at a maximum speed of six miles per hour
  • Equipped with automated driving technology, including software and hardware, enabling the operation of the device with the support and supervision of a remote personal delivery device operator

It also outlines rules for operating delivery robots including:

  • They must follow the existing rules of the road
  • They can only operate on sidewalks and crosswalks
  • There must be an operator who monitors and can take control of the robot
  • The robots themselves must be marked with the owner’s name and contact information
  • The robot must have brakes
  • The robot must have front and back lights for making deliveries from sunset to sunrise

With today’s signing, Washington becomes the eighth state to allow robot deliveries statewide, following Virginia, Idaho, Wisconsin, Florida, Ohio, Utah and Arizona. Robot delivery company Starship told us of today’s signing and said via email that it had worked with the state legislature to develop the regulations.

These types of robots are the perfect size for delivering meals (which. is. awesome.), but Seattle-ites probably won’t be able to get a burrito by ‘bot unless they’re a student at one of the local colleges. Companies like Starship and Kiwi have started out by deploying delivery bots to college and corporate campuses. Starship has programs running with George Mason University and Northern Arizona University. For its part, Kiwi just expanded to fifteen colleges across the country.

Now that personal delivery device regulations in place, one has to wonder if Amazon will ramp up its own robot ambitions. The company started testing out its Scout robot in the Seattle suburb of Sammamish in January this year, and the company has a patent for a delivery robot that would live at your house. Having a clear set of guidelines could give Amazon the go-ahead to go robot delivery wild in its home state.

Though I live in a suburb of Seattle, my town is pretty rural and hilly and sadly probably won’t get delivery robots anytime soon. All the more reason to make a trip to the big city, I guess.

If you want to know more about Starship and the complexities of building robots that deliver your food, check out this panel I moderated with Ryan Tuohy, Senior VP of Business Development at Starship, during our recent ArticulATE food robotics conference.

ArticulATE 2019: Robots on the Road. Automating Last Mile Delivery

Report: Food Robotics Market to Hit $3.1 Billion by 2025

One of the big reasons we put on our ArticulATE food robotics conference earlier this month, was that we could see the rise in automation throughout the food stack. Now Meticulous Research projects some numbers for that growth, with a report out today saying the size of the global food robotics market will hit $3.1 billion by 2025, growing at a CAGR of 12.7 percent between 2019 and 2025.

According to the press announcement, there are a number of factor that will drive this growth, including: “increasing food safety regulations, rising demand for advanced food packaging, growing demand to improve productivity, increasing production of low-cost robots, increase in investments for automated solutions in food industry, and growing demand for reducing production cost and increasing food shelf-life leading to increased adoption of robotic systems.”

Meticulous Research predicts that articulating robots (think: Flippy) will be the most popular form of food robot because their speed and accuracy continues to improve while their price keeps dropping.

Europe is the biggest user of food robots, followed by North America and Asia-Pacific. However, Meticulous Research predicts “significant growth” in food robot use in the Asia-Pacific during the forecast period because of investment in automation as well as growing demand for food safety regulation and packaged goods.

If you follow The Spoon then this report shouldn’t be a huge surprise. For the second year in a row, there were a record number of robots shipped in North America, with the food industry among the top non-automotive sectors buying.

More anecdotally, during our ArticulATE conference, we heard from companies across the food world talking about how a growing global population and its desire to, you know, eat, combined with a labor shortage is generating increase interest automation. As this report notes, these food robots are getting better (Flippy is on its third generation), smaller and cheaper. And from my personal experience, robots are pretty great at making food as well.

Video: Google Brain Director on Creating Robots for a Messy World (Like Kitchens!)

Robots are really good at moving around. They can roam the aisles of grocery stores easily or manipulate an articulating arm to pour you a cup of coffee. But what robots aren’t good at is understanding the world around them.

“The human world is by definition messy and complicated,” Vincent Vanhoucke, Principal Scientist and Director of Perception and Robotics for the Google Brain Team told the crowd during his presentation at our recent ArticulATE food robot conference in San Francisco.

Vanhoucke’s team is working on taking the things robots do well — moving around — and marrying that with advancements in computer vision and deep learning to make robots more useful in the messy and complicated real world. And it turns out that food in particular, with its different textures and properties, is quite messy and complicated.

The Google team does this by training robots to pick up objects of various sizes and shapes up using deep learning. Through recognition, repetition and reinforcement, the robots develop their own strategies and behavior for solving problems (the inability to pick up an object) on their own without a human programming specific solutions.

The applications for this can be seen in something like a feeding robot for the disabled. Rather than having a “dumb” arm that only scoops food from a predefined area in a bowl and lifting that food to a predetermined height, a deep-learning enabled robot can identify different food on a plate no matter where it is, pick it up and lift it directly to a person’s mouth.

It’s really fascinating and cutting-edge science, and you should definitely watch the video of his full presentation.

ArticulATE 2019: Using Robotics in Messy Environments (Like Kitchens!)

Be on the lookout for more videos from ArticulATE 2019, coming throughout the week.

Take an Instagram Food Robot Tour of San Francisco! Cafe X, Yo-Kai Express, Creator, Blendid and Le Bread Xpress

One of the reasons we hosted our ArticulATE food robot conference in San Francisco this week is because, well, it’s where most food robots are. So it only made sense while I was in town to go on a Food Network-esque trip around town, enjoying all the robot repasts I could.

My tour did not disappoint. I spent the day walking (and Ubering) around SF visiting Cafe X, Yo-Kai Express, Creator, Blendid and Le Bread Xpress. Because it’s better to show than to tell, we created a Spoon Instagram (follow us!) to give you a glimpse at all the cool food robotics happening right now.