Researchers will actively test machine learning models in the field this spring using the latest innovation from the TerraByte project led by Dr. Christopher Bidinosti, Department of Physics, and Dr. Christopher Henry, Department of Applied Computing, University from Winnipeg.
It shows what the future could hold for robotics and autonomous vehicles in agriculture.
Dr. Christophe Bidinosti
Along with postdoctoral researcher Dr. Michael Beck, they recently partnered with R-Tech Industries to design and build the Data Rover, a lightweight, solar-powered, remote-controlled machine that will roam through soybean plots on large bicycle tires to collect large volumes of crop data as digital images as well as data from other sensors for hyperspectral and 3D imaging.
Bidinosti was thrilled with how quickly the R-Tech Industries team brought UWinnipeg’s vision to life.
“We brought them some ideas and within a few months they had it built, which is really remarkable,” Bidinosti said. “It shows what the future could hold for robotics and autonomous vehicles in agriculture.”
R-Tech Industries specializes in custom agricultural research equipment, working with researchers to create innovative machines.
“We listened to Chris and his team to understand what the goals of the project were, drawing on our experience to make suggestions for the final design,” said Rob McClement, owner of R-Tech Industries. “As this was a unique project and somewhat outside of our area of expertise, we worked with Shea Hunt of Valley Prototyping Solutions to design the drive systems. R-Tech managing director Liam Riddell designed the frame and mechanical systems and incorporated Shea’s work into the completed project.
Daryl Domitruk, executive director of Manitoba Pulse and Soybean Growers (MPSG), connected Bidinosti and Henry with R-Tech Industries.
Domitruk provided the research team with essential information about Manitoba’s agricultural sector and presented it to local farmers whose fields were used for data collection.
“Digital agricultural applications such as artificial intelligence and machine learning are potentially game changers,” he said. “However, there needs to be experience-based knowledge transfer and a lot of peer-to-peer learning among farmers before they can fully exploit this potential. It is complicated. UWinnipeg researchers give us the opportunity to understand this technology objectively. Our role is to explain it to farmers and remove part of the mystery. This is essential for farmers to eventually adopt these technologies.
Digital agriculture helps farmers grow more resilient crops
The Rover is the realization of Bidinosti and Henry’s shared vision to create a database of high-quality tagged images accessible to industry and academia.
Now that the Rover is built, the research team will continue working with MPSG and NorthStar Robotics to develop software and conduct field trials during the 2022 growing season, supported by a grant from Ag Action Manitoba through the Partnership Canada for Agriculture, as well as by Mitacs Accelerate Grant with MPSG as an industrial partner.
To achieve this vision, Henry and Bidinosti spent many hours building relationships and cultivating partnerships. These interdisciplinary connections have been essential to the continued success of the project.
“We have the expertise in data collection and machine learning sensors. R-tech builds specialized agricultural equipment. MPSG gives us insight into the range of agricultural production problems that machine learning can solve. NorthStar Robotics specializes in autonomous vehicle control, which will allow us to test machine learning models in the field and collect more data,” said Henry.
UWinnipeg graduate Alex Krosney is a member of the TerraByte research team. He is currently working with NorthStar Robotics on the interface between the Rover’s Jetson (brain) and its microcontroller.
“The Jetson sends commands to a microcontroller and communicates with the rover’s motors so it can move autonomously. From an anatomical point of view, it can be analogous to a kind of nervous system. Once we’ve completed the bridge between the Jetson and the microcontroller, I’ll move on more to the “brain” side of things,” Krosney said.
This technology will eventually allow farmers to care for individual plants in commercial fields the same way we care for plants in home gardens.
“Of course, this is a massive task that requires a significant level of automation,” Krosney said. “The rover will be used to collect a large number of images of plants in an outdoor setting. These images will then be used to train neural networks to perform machine learning tasks such as plant detection, classification cash, health assessments, etc.
The key collaboration for the success of the project
When they first announced their research in 2019, the vision of automatically generating and labeling images through a computer-controlled camera system was moot. By collaborating with industry partners with diverse resources and expertise, Bidinosti and Henry are able to push the boundaries of what is possible in digital agriculture.
“This will improve crop breeding, crop disease research and organic farming practices to ultimately provide farmers with more resilient and drought-tolerant crops,” Bidinosti said.
In 2019, UWinnipeg, in partnership with EMILI, received $2.4 million from Western Economic Diversification and $250,000 from a Weston Seeding Food Innovation grant. Since then, they have received significant funding from Mitacs and they continue to develop TerraByte with industry and academic partnerships across the country.
“It takes all of us working together to achieve this vision,” Henry said.