CHAMPAIGN, Ill. – Robots are being developed and studied to combat resistant weeds in agriculture.

Madhu Khanna, professor of agricultural and consumer economics in the College of Agricultural, Consumer, and Environmental Sciences and director of the Institute for Sustainability, Energy, and Environment at the University of Illinois Urbana-Champaign, said there are a couple of reasons why robots are being investigated as a way to control resistant weeds.

One reason is that weeds are becoming resistant to glyphosate due to overapplication, and it’s difficult to develop other chemical pathways to overcome this resistance. Additionally, in current herbicide management practices, herbicide is applied before plant emergence or up until the canopy of the plant closes. Once the canopy closes, it’s very difficult to get to the weeds.

“The robots, for one, are able to do mechanical weeding. They don’t rely on chemicals, so they can eventually get rid of weeds and reduce resistance over time,” she said. “The robots can also go under the canopy of the crop while it’s standing, so you can deploy them multiple times during growing season. This allows for multiple rounds of weed removal throughout the season, which should hopefully help yield.”

Taking on superweeds is just the latest iteration of what agricultural robots can do for farmers. Previously, attention has been focused on the TerraSentia, a robot developed by researchers at the University of Illinois to generate a detailed portrait of an agricultural field. Using a camera and a lot of computer programming, TerraSentia can determine the size and health of plants, measuring plant height, stem diameter, and leaf-area index. In corn fields, the robot’s data can tell growers the number and quality of ears each corn plant will produce.

Similarly to TerraSentia, the robotic weeder uses a camera to browse plants under the canopy. Through computer vision, the weeder can differentiate crop plants from weeds and remove unwanted weeds by pulling hoes through the soil, disturbing the emergence of weed seeds. Khanna said that in Europe, researchers are working to improve robotic weeders’ capacity to distinguish between different types of weeds so that beneficial weeds aren’t removed. The robots are small, lightweight, and are being designed to be rechargeable.

Khanna and her team studied the factors influencing the adoption of robotic weeders in agriculture. The study focused on controlling common waterhemp in corn crops. Khanna said farmers face several decisions when considering robotic weeders, such as how many robots to adopt, what percentage of the field should be managed robotically rather than with herbicides, and how frequently the robots should be deployed in the field.

“There’s this balance between the cost of the robots versus the damage that would be caused by the weeds to the yield and how soon that damage is going to occur,” she said.

The adoption of robotic weeders over traditional weed control methods comes down to management style. Hypothetically, in the first six years of a robotic weeder being commercially available, farmers who make weed control management decisions based on one year at a time would not use robots at all. Instead, they would likely continue to apply herbicides until they are no longer effective and then shift to robotic control when chemical options are exhausted. This may be the most likely scenario for tenant farmers.

On the other hand, farmers who use forward-looking management styles are likely to adopt robots much earlier if they were available. In her calculations, Khanna found that assuming a robot costs $20,000, farmers with forward-looking management perspective are likely to adopt robots if 0.0001 percent of weed seeds are resistant to chemical applications. For those making a decision year by year, that percentage jumps to 5 percent.

Khanna’s research concludes that it is optimal to adopt weeding robots before weed susceptibility is fully exhausted regardless of management strategy.

“Forward-looking farmers would actually be able to manage (weeds) with fewer robots, because they would be strategically able to combine herbicides and robotic usage to manage their resistance,” she said. “They would be able to delay resistance at a lower cost than those farmers that would wait to adopt until the problem has become intense and costly.”

The weeding robots being studied at the University of Illinois are not commercially available. However, other weeding robots are on the market, typically for specialty crops that are low to the ground so that the weeder can go over the tops of the canopies.

While Khanna’s research is hypothetical, it’s an important first step in understanding what it will take for the widespread adoption of weeding robots. Future research should focus on real-world applications, such as talking to farmers to understand their opinions on the practice.