Skip to Content

Robots Start to Grasp Food Processing

Advances in robotics make it possible to automate tasks such as processing poultry and vegetables.

It is less striking than Deep Blue’s victory over chess champ Garry Kasparov, but Richard van der Linde says that his robotic hand’s mastery at picking up cabbage is something of a milestone for machines. With the aid of five cameras, plus sensors in its wrist to monitor the resistance it encounters, the three-fingered gripper can carefully pick up a cabbage, reorient it, and place it into a machine that removes the core. “In industry, only humans can do that at the moment,” says van der Linde.

His company Lacquey, based in Delft, the Netherlands, is working with FTNON, a manufacturer of food-processing equipment, to get the technology ready to go to work inside the giant chillers where today humans process cabbage, lettuce, and other produce for packaging. ­Lacquey is also testing versions for other sorts of jobs, such as packaging tomatoes, peppers, and mangoes.

The company’s progress is an example of how advances in robotic manipulation technology are opening up new jobs for robots in the food-processing business. Solid, hard, identical objects such as car parts are easy for robots to move around. But delicate, flexible, naturally variable objects such as meat, fruit, and vegetables require much more sophisticated sensing and manipulation.

Interest is driven partly by the potential to cut labor costs, just as in other industries. But food-processing companies also see robotics as a way to increase safety, says Gary McMurray, who leads the Food Processing Technology division at the Georgia Tech Research Institute. “Anywhere you have people in there handling food, they make mistakes from time to time,” he says. Incidents where meat or vegetables become contaminated with, say, E. coli or Listeria are costly to a food processor. A 2015 study found that on average, meat recalls wiped $109 million from a public food-processing company’s value within five days of their announcement. Though figures are not available for the specific number of cases originating from contamination at a food-­processing plant, the Centers for Disease Control estimates that 128,000 Americans are hospitalized with food-borne illness from all causes each year, and of those, 3,000 die.

McMurray’s research group is currently developing two systems for the poultry industry. One can grasp a chicken carcass moving along a production line and cut the shoulder tendons in preparation for the removal of the breasts and wings. That system can already match the average yield of a human worker. In a second project, a low-cost two-armed robot called Baxter, produced by Rethink Robotics and designed to work safely alongside humans, is being programmed to place poultry carcasses onto the cone-shaped holders that carry them through a processing plant.

Both systems rely on visual and physical feedback. For example, the cutting robot uses a 3-D vision system to estimate the location of a chicken’s joints and tendons. It then uses sensors on its knife to “feel” whether it is cutting meat or bone. “Working with these wet, deformable, slippery objects is challenging, but it seems to be doable,” says McMurray.

Getting a robot to do a task like that well usually requires engineers to carefully program in specific techniques and commands. But machine-learning software could automate much of that process and make it practical for robots to carry out more complex tasks with a variety of foods, says Ashutosh Saxena, an assistant professor of computer science at Cornell University.

He used that approach to teach a two-armed robot to assemble a simple salad. The robot first went through a training phase in which it used knives, spatulas, forks, and other implements to probe the physical properties of foods including tomatoes, lettuce, and cheddar cheese. Afterward, the robot could figure out for itself how to slice up the elements of the salad and then move them around.

It is unclear when, if ever, it might be possible for a robot to keep pace with a human chef. In the near term, van der Linde says, Lacquey and others have to prove that their machines can match or exceed the pace of humans doing specific tasks on existing production lines.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.