Researchers use machine learning to boost basil flavor

0
376
MIT grows tasty basil with machine learning

If you’re looking for a way to punch up your pesto sauce, MIT researchers have presented a hi-tech, flavor-boosting solution.

In a recent study, a group of scientists from MIT’s Media Lab Open Agriculture (OpenAg) group and the University of Texas at Austin used machine learning to grow tastier basil plants. In this effort, they used computer algorithms to create growth conditions which would optimize flavor.

PLOS ONE published the results in an April 5 report. The bottom line: “Climate recipes” created in the experiment boosted basil plant flavor.

Advertisement
Manage your supply chain from home with Sourcengine

Does this mean a new era of computerized farming awaits? The study group hopes so.

Changing the Norm

Basil grows outdoors in the summer or indoors year-round in a container or windowsill herb garden. In either environment, plants require six to eight hours of sunlight a day to grow.

In this study, the MIT team aimed to create conditions which would help them discover and implement a growth plan to produce more flavorful basil. To achieve this goal, the team bucked traditional methods and grew thousands of plants inside containers, varying the color, intensity, and duration of light.

Moreover, via this hydroponic, vertical farming method, conditions like light, water, nutrients, temperature, and other variables are artificially generated under computer control.

Overall, adjusting these factors would be impossible in a natural setting. So, scientists had to create the controlled environment to achieve their desired outcome.

Surprising Discovery

In the experiment, MIT’s innovative researchers grew plants inside containers they called “personal food computers.”

Furthermore, they adjusted variables such as light, soil type and watering frequency, collecting data as plants grew. Once grown, scientists used gas chromatography and mass spectrometry to measure the concentration of flavor molecules in the leaves. In doing this, they also evaluated the taste of the basil.

“We’re really interested in building networked tools that can take a plant’s experience, its phenotype, the set of stresses it encounters, and its genetics, and digitize that to allow us to understand the plant-environment interaction,” said Caleb Harper, director of MIT’s OpenAg Initiative and MIT Media Lab principal research scientist in a press release.

The team ultimately found exposing plants to 24 hours of light each day yielded the best flavor, which was surprising. They couldn’t have gained this information in a traditional agricultural setting, study co-author John de la Parra explained.

“You couldn’t have discovered this any other way. Unless you’re in Antarctica, there isn’t a 24-hour photoperiod to test in the real world,” he said. “You had to have artificial circumstances in order to discover that.”

Computerized “Climate Recipe”

When researchers fed all experimental data into machine-learning algorithms developed by MIT and Cognizant teams, the algorithms generated “climate recipes” to maximize flavor. Ultimately, the study concluded the computerized method “can find growth recipes that are both effective and surprising—and difficult and time-consuming to find through traditional hand-designed experiments.”

Data Sharing

MIT isn’t alone in researching and developing hi-tech hydroponic farming. For instance, China is using rooftop hydroponic farming models and IKEA is teaming with BONBIO to grow and harvest lettuce in Sweden. Unfortunately, some companies choose not to share technology.

MIT scientists, however, are making their data open and free to the public. This makes it easier for other scientists, farmers, and non-profits to join a common urban farming initiative.

“Our tools being open-source, hopefully they will get spread faster and create the ability to do networked science together,” Harper said.

In a widespread educational initiative, MIT sent “personal food computers” to middle school and high school students across the United States. Overall, the global network of users spans 65 countries.

Box recipients can grow plants in their own controlled environments and share ideas and results in an online forum.

More importantly, they can send the data back to MIT. This means as students monitor progress and report their findings, the invaluable educational opportunity is a win-win for both sides.

Other Uses and Future Implications

MIT researchers are moving beyond flavor enhancement and are working to develop basil plants with higher levels of disease-fighting compounds. For example, some basil compounds help control blood sugar.

In prior work, de la Parra showed varying environmental conditions on plants effectively boosted these compounds. Today, researchers hope achieving this goal might be helpful in combating diabetes.

Additionally, the team is considering using the approach to increase yields from medicinal plants. They also plan to study how cyber agriculture can help climate change adaptation.

“You can see this paper as the opening shot for many different things that can be applied, and it’s an exhibition of the power of the tools that we’ve built so far,” de la Parra said. “This was the archetype for what we can now do on a bigger scale.”

Overall, this eye-opening MIT study is only the beginning. As research shows, leveraging machine-learning in controlled agricultural environments could yield all sorts of amazing results which could impact the future of farming and beyond.