A researcher, along with industry partners, is giving the age-old task of combating weeds in agriculture a smart technology upgrade.
Medhat Moussa, a professor in electrical and computer engineering at the University of Guelph, has developed a tractor attachment that utilizes computer vision and artificial intelligence (AI) algorithms to identify and manage weeds in lima bean crops.
It then creates maps that enable farmers to spray herbicides with greater precision and efficiency. The technology aims to save farmers time and money by reducing costs and minimizing the amount of chemicals wasted.
Moussa told The Observer that weeds were a big problem for farmers of different crops, and in the case of lima beans, they also affect the harvesting of them and the yield.
“The current practice is to spray so that you mitigate and you try to reduce the pressure from weeds. What type of spray pesticides you’re going to use, and the rate of spraying, is dependent on what the farmers feel the weed pressure in the field is.”
To do this, they have to go to each of the fields and scout areas manually. On a 100-acre farm, they are counting perhaps one per cent of the field to decide how much and where they should spray the herbicide, he explained.
“Our robotic attachment, equipped with cameras, gets mounted onto a small tractor or similar vehicles. As the tractor is driven through the field, the cameras scan the ground and take photos that are analyzed by a three-stage AI algorithm. Our algorithm differentiates between lima beans and weeds in each image, then combines the images into a complete density map showing exactly which areas need to be treated.”
These pictures are then processed using AI to visualize the exact weed pressure in every area of the field and create a very precise weed density map. This map can then be uploaded to the herbicide sprayer for a more precise dispersal of spray.
“On that basis, when you are going and spraying, you can adjust the spray depending on the area. You create a weed density map that you upload into the sprayer to adjust accordingly,” Moussa explained.
The process will not only make the current year’s harvest more efficient, but also provide many benefits in future years, as well.
“What happened is, you go this year and spray, and have information about the weed density that you started with and what happened at the end of the season and so on.”
“Next year, you have a very good understanding of every corner of that field, and that will enable you to adjust your practices, hopefully leading to higher yields and less cost.”
Moussa worked with Nortera Foods, a fruit and vegetable producer and end-user of the system, and Haggerty AgRobotics after they presented him with the challenge of improving spraying for their lima bean crops.
“Our lab has been working in the automated agriculture space for years using technologies such as computer vision and machine learning,” he said.
“It’s a great fit because our partners have the capabilities to test our technology on a larger scale, show it to farmers and bridge the gap to market adoption.”
He noted that robotics technology in agriculture tends to have a tough time being adopted by farmers, especially those who promote full automation. That is why he believes that his combination of automation and traditional farming practices will make it more appealing for farmers to adopt.
“The issue about autonomy here is that, as a researcher, you think absolutely, ‘Yeah, let’s go fully autonomous,’ stuff like that. However, I think that a middle ground, now that might be better existing machinery more effective and utilize it in a more optimized way.”
Many of these new technologies aim to reduce operational costs by cutting labour, but that’s not the case in this instance.
“You want to have better, sustainable agriculture, but you’re not necessarily wanting to really go and fire all of these people. That’s not the aim of this project.”
There are other autonomous weed removal units in Europe and the United States, but they have not been commercially successful due to their slow operation or high cost. That is why Moussa’s project team wanted to improve and upgrade the current practices without having a negative impact in terms of labour.
The product is in the final stages of testing and has the potential to be used for other crops in the future.
“I think that it is better to reach that point where we introduce a technology to optimize current operation without having any negative impact in terms of the labour, of course.”
