Technologies That Are Reinventing the Food Industry

Shape-shifting noodles and an AI system that can give recipes for meals you’ve photographed

16 August 2017

Have you ever wished you could replicate a delicious restaurant meal but couldn’t identify the ingredients? Or had a late-night craving for pizza and realized that all the pizzerias near you were closed?

Engineers, including a few IEEE members, are working on technologies designed to change how we package, obtain, and prepare food.

  • SHAPE-SHIFTING NOODLES

    MIT’s Tangible Media Group has developed noodles in the form of flat sheets of gelatin and starch that take shape after being exposed to water. The noodle sheets can be stored in smaller packages, taking up less shelving space and costing less to ship. The group works to incorporate technology into everyday tasks including meal preparation.

    The noodles can take the shape of common pasta, like macaroni and rotini, or even unconventional forms like flowers, according to MIT News.

    The researchers first determined how to control where and to what degree the structure of gelatin and starch bends to create different shapes from the noodles. They then used a 3D printer to build strips of edible cellulose over the top gelatin layer, so the strips could act as a barrier to absorb the right amount of water.

    IEEE Affiliate Member Hiroshi Ishii, an MIT professor of media arts and sciences, and IEEE Member Daniel Levine, a graduate student, are on the research team.

  • FAST-FOOD PIZZA

    Vending machines that churn out pizza pies could be the next fast-food fad. The first 24-hour Pizza ATM in the United States can be found at Xavier University in Cincinnati. When students want pizza late at night, they can order it instantly from a machine.

    The company behind the invention is Paline of Lemieux, France. Its Pizza ATMs have been place in hundreds of locations, mostly in small pizza stores and independent businesses throughout Europe.

    Paline is partnering with TSS Technologies of Ohio for its U.S.-based vending machines. The Pizza ATMs are equipped with refrigeration and an electric convection oven.

    Owners place pizzas in special boxes in cold storage inside the machine, which can hold 70 pies of up to 30 centimeters in diameter. Customers select which toppings they want from a touch screen, specify whether they would like the pizza served hot or cold, and then swipe a debit or credit card to pay. Each pie costs US $10.

    The machine costs $55,000. Businesses that purchase it need to have someone prepare and pre-cook the crust and add toppings.

    For those who want to own a pizza vending machine but don’t have a staff to make the pies, Paline is offering Pizza Chef School courses in Cincinnati.

    Potentially, a large pizza company could set up Pizza ATMs outside its restaurants after hours, or place them where it doesn’t have a pizzeria.

  • RECIPE IDENTIFICATION

    MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has trained an AI system, Pic2Recipe, to analyze photos of meals, identify their ingredients, and suggest similar recipes.

    IEEE Member Yusuf Aytar told MIT News that it has been difficult to analyze food using computer vision up to now, because there haven’t been enough images to work with, but Instagram and similar services are changing that.

    “Seemingly useless photos on social media can actually provide valuable insight into health habits and dietary preferences,” Aytar says.

    He wrote a research article on Pic2Recipe with IEEE Affiliate Member Antonio Torralba, an MIT professor of computer science and AI.

    Using Recipe1M, a database of more than 1 million recipes from Allrecipes.com and Food.com, the team trained a neural network to find patterns and make connections among images and recipes.

    The researchers integrated data from Food-101, a joint project of the University of ETH Zurich and the University of Leuven, in Belgium, which developed an algorithm to recognize food images with a 50 percent accuracy rate. The data from Food-101 and Recipe1M were combined to create Pic2Recipe.

    The Pic2Recipe demo on CSAIL’s website has been able to identify the exact ingredients in some uploaded food images and to suggest recipes that it determined to be similar.

    The platform recognizes ingredients down to flour and butter as well as the composition of, say, muffins. As of now, it has a harder time with ambiguous foods such as smoothies.

Learn More