Artificial Intelligence and Machine Learning News

Adding Smarts to Vending Machines Drives Convenience, Efficiency

Advanced feature sets and functionality are projected to drive the market for connected vending to nearly nine million units by 2024.
man using a smart vending machine, illustration
  1. Introduction
  2. Real-Time Monitoring, Payments, Control
  3. Expanding Use Cases with Climate Control
  4. Smart Vending in Retail
  5. Leveraging AI to Enhance Personalization
  6. Author
man using a smart vending machine, illustration

Vending machines, which allow people to easily purchase items without interacting with a human worker, have been around since the 1st century, when a Greek engineer and mathematician named Hero Alexandria created a machine that accepted a coin before dispensing holy water at a temple, to prevent people from taking more than their share of holy water.

Two millennia later, a far greater number and variety of products can be purchased from vending machines, thanks in part to the advent of new technologies including always-on, Internet of Things (IoT) connectivity, advanced physical and digital controls that allow these machines to be placed in a wide variety of settings, and the use of artificial intelligence (AI)-based algorithms that can capture and analyze customer insights, improve stocking efficiency, and deliver greater levels of personalization to customers.

The global installed base of connected vending machines reached an estimated 2.4 million units in 2019, according to Berg Insight, a research firm that tracks the installed base of connected vending machines. Connected vending machines are equipped with an always-on Internet connection, which allows data to be sent between machines in the field and management software, enabling real-time payments, monitoring, and remote management of the machines.

Advanced feature sets and functionality are projected to drive the market to nearly nine million units by 2024, according to Berg Insight, helped along by the desire of organizations to better serve customers without needing to attract and retain relatively costly human workers.

Figure. Technology has allowed the humble vending machine to sell an increasingly diverse set of products. An example shown here is a Smart for Two vending machine in Shenyang, China.

Back to Top

Real-Time Monitoring, Payments, Control

The most advanced smart vending machines deployed today leverage wireless IoT connections that support several back-end functions, including real-time card-based and contactless payments, the ability to connect with mobile apps for product ordering and in support of age-verification tools, and the use of algorithms that analyze past behaviors to provide personalized recommendations. In some countries that have a strong history of vending machines being deployed outside of secure locations, smart vending machines can provide an added layer of security and convenience, a result of their ability to accept cashless payments.

For example, in Dubai, smart vending machines have been deployed to provide freshly baked bread for free to needy citizens. The “Bread for All” vending machines have a computer touchscreen that allows people to select from three different types of bread, then to press a button to receive loaves for sandwiches, pita bread, or flat Indian-style chapatis. An IoT connection is used to accept credit cards so other citizens can donate money to help fund the program.

Smart vending machines can also be used to dispense products that require special conditions, such as a constant temperature, as well as ensuring products not safe for human consumption will not be dispensed. Farmer’s Fridge has created a network of 400 smart refrigerators placed in airports, hospitals, office buildings, and universities around the U.S.; they incorporate technologies designed to help people access fresh foods on the go. Users can easily browse a touchscreen to read full nutrition information on the items being vended, while a built-in thermometer scans the unit’s internal temperature every five minutes. If the electrical power is interrupted for even a few minutes, the fridge will deploy a lockout function that prevents anything from being sold. Connected to a management software application located on a centralized server via a 5G IoT wireless connection, the fridge also incorporates an algorithm that minimizes food waste by learning the purchasing preferences of customers so the right mix of products can be delivered to the fridge every time.

Back to Top

Expanding Use Cases with Climate Control

Smart vending machines are also using temperature management to provide greater flexibility, in terms of the products that can be vended or stored inside them. Smart storage lockers can be compartmentalized so that one area can be used to keep items hot, while another portion can be kept cool or even frozen. This variable temperature control allows for maximum flexibility, and can support several use-cases.

David Ashforth, co-founder and partner of Digital Media Vending International, LLC, a custom vending machine designer and manufacturer, cites his company’s development of a smart vending machine to sell cup-cakes. The machine is equipped with temperature controls to keep the food fresh and appealing; it also can support digital payments, and automatically remains in contact with the business so it is alerted when the machine needs to be restocked or otherwise serviced. This use-case may be especially valuable for small or medium-sized businesses that want to expand production of food products, but are limited by the number of workers they can hire to staff additional retail locations.

“A lot of small companies are doing the math and thinking, ‘Well, I can spend the $15,000 on a vending machine and it’s still cheaper than a year’s worth of wages’,” Ashforth says. He adds that such smart vending machines can also be configured as climate-controlled food-delivery storage lockers. These could be placed in office buildings, apartment buildings, college dorm lobbies, or other locations where security controls impact the efficiency of food deliveries. Using a digital payment and unlocking code, customers could order food via an app, and then the delivery could be made to a locker in a common space, which could be only opened via a code or payment mechanism. The ability to select the right temperature for a specific delivery (heat, cool, or frozen) improves the quality of the experience for the customer, while improving the efficiency of delivery drivers by providing a safe, convenient drop-off location, Ashford says.

“I can spend the $15,000 on a vending machine and it’s still cheaper than a year’s worth of wages.”

That model is being flipped in Italy, where software company Vending Automatic Delivery Operations (VADO) introduced an e-commerce solution that allows vending machines to sell products through delivery apps such as UberEats, Deliveroo, Glovo, and Just Eat. Instead of forcing delivery people to visit a store and wait for an order to be filled, VADO creates a menu in the delivery apps for products in a connected vending machine, which can be ordered by a customer. Then a delivery person receives the order, travels to the machine, and uses the order code to prompt the machine to dispense the order and bag, which the delivery person then takes to the customer. Currently, VADO’s vending machines are deployed in the cities of Turin, Milan, Rome, and Salerno, where they offer late-night snacks, ready-to-eat prepackaged foods, drinks, and OTC medical products.

Back to Top

Smart Vending in Retail

Smart vending machines are also projected to become popular within traditional retail stores, particularly if they “improve the customer experience, and don’t burden the in-store [labor] team that’s already stretched,” says Rebecca Faulconer, president and founder of Retail Automation Consultants, which has worked with Best Buy, Procter & Gamble, and Rite-Aid to provide smart vending solutions. According to Faulconer, her firm’s “conversations with [companies] are around store design or strategy, and the thought process is more about solving a [specific] problem.”

Smart vending machines are also being used to manage operational issues caused by a lack of workers, or the rising costs of using human workers to handle products that are prone to shrinkage (an industry term for stealing). “Last week, I was in New York City and I went into a Walgreens and literally everything was locked up,” Ashforth says. He explains that he found relatively low-value merchandise like toothpaste and shampoo locked up, requiring him to press a button to summon a sales associate to unlock the case containing the items he wanted to purchase, a process that increases customer frustration while taking employees away from higher-value tasks.

“My vision forward is automating those shelves,” Ashforth says, noting the machines his company produces can be as large as 32 feet wide, mimicking the footprint of many retail store shelves. “I feel like we are really creating the future of retail, in the sense that these big box stores won’t need to lock away their product any more, and are making sure consumers have instant access to that product instead of needing to find a clerk to unlock it.”

Automated retail vending solutions such as these can also be used within stores to handle age-restricted products, ranging from certain medications to alcohol, or even recently legalized cannabis products. The big hurdle with using vending machines for such products in the past was the lack of a secure age-verification solution that could easily be deployed, says Teudis (Teddy) Sanchez, co-founder and chief technology officer of Optimal Station, a Florida-based custom integrator of automated retail hardware and software solutions.

Sanchez says his company developed vending applications that allows customers to purchase age-restricted products via an automated vending machine. Customers select such an age-restricted product, pay via cash or other standard payment method, and before the machine dispenses the product, an AI-powered facial recognition camera captures an image of the purchaser’s face. The purchaser then inserts a government-issued photo-bearing identification card into the machine, which then compares that image with the one just captured by the machine, to ensure a facial match.

“I feel like we are really creating the future of retail, in the sense that these big box stores won’t need to lock away their product any more.”

“It authenticates the document via hologram scanning,” Sanchez explains, noting there needs to be an exact match based on consistent features, such as pupillary distance or other facial markers that don’t change if a person ages or otherwise changes their appearance. “If there’s any doubt that it’s not the right person, the machine will reject the purchase, and allow the person to try again, or get their money back.”

Sanchez says smart vending machines are a more consistent way of accurately verifying a person’s age, as well as ensuring that accurate payment is made for products. However, he says, such automated solutions likely will still be located within a retail store, where secondary measures can be put in place (such as requiring prospective purchasers of marijuana products to sign in), as well as offering better physical security to prevent vandalism or theft of products.

Back to Top

Leveraging AI to Enhance Personalization

Perhaps the most innovative uses of smart vending machines are being deployed in markets where consumers are more accustomed to vending machine technology, such as Japan, which has embraced vending technology for years, and whose population is a bit more comfortable with cutting-edge technology such as facial scanning being used in a commercial setting.

In support of its KATE makeup brand, Tokyo-based Kanebo Cosmetics created an automated retail experience called the KATE iCON box. It uses a camera installed on a vending machine to capture an image of a customer’s face, which an AI algorithm uses to conduct a facial analysis that is the basis for suggesting a palette of four personalized eye shadow colors. The machine prints a sheet with the palette, which can then be taken to a sales associate to purchase the suggested eye shadows. This approach makes the selection process more convenient for customers, and drives greater interaction with the brand without requiring costly human labor to go through the make-up assessment process with the client.

Perhaps the greatest benefit to adding intelligence to vending machines is that it allows companies that have primarily interacted with customers online to provide a way to not only make their products available to people in the real world in a convenient format, but also to extend their brands via an interactive experience.

“The automatic retail channel in general gives the ability to brands to have a physical presence without the expenses of a brick-and-mortar store,” Sanchez says. “So we’re really seeing companies that have a huge online presence using these smart vending machines like pop-up shops. And it helps make the instant connection with the user that was always getting ads, but were never clicking on them.

“Now, the customer is like, ‘Oh, this is the company that was always sending me ads, but look, it’s physically here, so let me go check it out’.”

*  Further Reading

Connected Vending Machines. Berg Insight, March 2020, Fourth Edition;

How Smart Vending Machines Work;

Smart Vending Demo:

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More