Weed Vision: How Augmented Reality Can Improve Grow Houses

Imagine a future where farmers wander through the rows of a greenhouse, wearing augmented reality glasses that tell them what type of tomato or strain of marijuana they're looking at. Cameras installed above watch for signs of disease and pests, and monitor watering needs and growth rates. If the system spots an anomaly, it pings one of the augmented farmers to isolate the problem before it spreads.

Huxley, a system of augmented reality and artificial intelligence for produce greenhouses and cannabis grow-operations, is developing what founder Ryan Hooks calls "plant vision."

Some farms already use smart tags and data-collection drones to optimize their production. There's an app for optimized irrigation and researchers help farmers harvest crops like they're lines of code. But what Hooks wants to do is enhance the farmers themselves, to work alongside artificially intelligent machines with augmented reality.

"What I learned trying to invent in this space, was that nobody had done augmented reality yet for greenhouses," Hooks told me. There are plenty of apps for greenhouse growers, but none are hands-free—a function that's important and obvious, he said. You need your hands to get dirty. Wearing a passively augmented device, such as glasses, could guide a farmer with a rich amount of data without getting in the way.

Augmented reality vertical farming could help a small farm scale more quickly. It's hard to find affordable labor for a futuristic farm: People who know how to balance PH, humidity, and climate, and who are also willing to work for a small-operation paycheck. "As small farms scale they can't find enough people for the job, limiting their growth potential," Hooks said. "If someone can drive Uber, then they can put on Huxley and become a modern farmer."

While these augmented farmers work, they'll help build a global dataset that will train the AI on thousands of varieties of fruits, vegetables, and marijuana plants (and the unique diseases that go with each. Hooks plans to partner with universities and researchers to build the database.

The process of training the machine vision that goes into Huxley could serve as research and development for an automated future, where AI-guided machines browse greenhouses, pick ripe produce, and nip diseases in the bud on their own. It could make good food and medicinal plants more affordable and accessible, Hooks said. "This scenario allows for enhanced labor, bringing more efficient, pesticide free plants to the masses."

Plant vision is a potential boon for the cannabis industry, as well. Since cannabis is still illegal on the federal level, the EPA hasn't set guidelines for pesticide use on crops, and it's up to states to test for pesticide levels on their own and set maximal limits. Some cannabis products in Washington, Colorado and Canada have been recalled for pesticide levels found in the plants. If growers are able to scan for disease and keep pests in check without chemicals, they could spare the entire crop from chemicals. For a small grow operation, avoiding pesticides eliminates that risk.

Subscribe to Science Solved It, Motherboard's new show about the greatest mysteries that were solved by science.