A team of Cornell University undergraduates has won the grand prize at the 2026 Farm Robotics Challenge with an autonomous robot that kills weeds using small electric shocks rather than chemicals. Drawn from the College of Agriculture and Life Sciences and the College of Engineering, the students beat out a field of 96 teams from 13 countries to take the top award and a $50,000 prize, which they have used to launch a company called Rootline Robotics to keep developing the machine.
The robot tackles a problem that has frustrated growers for years, which is how to control weeds close to tree trunks and grapevines without spraying herbicides or relying on scarce manual labor. Instead of chemicals, the system delivers low-energy pulsed high-voltage microshocks that disrupt a weed at the stem and root without the high power draw that has held back earlier electric weeding tools. The approach gives orchards and vineyards a chemical-free option for the precise, plant-by-plant weeding that is hardest to automate.
What makes the machine work is the way it sees and reaches each weed. The robot pairs computer vision, machine learning, and depth sensing to identify weeds and judge their position in three dimensions, then uses a two-degree-of-freedom mechanical arm tipped with a comb-based electrode array to make contact. The comb conditions the plant’s geometry before the shock, lining up the stem so the pulse lands where it does the most good. The result is a system that can pick out a single weed next to a trunk and treat it without touching the crop.
The team built the robot on the Amiga platform from Bonsai Robotics, the wheeled base that the Farm Robotics Challenge provides as a common foundation, and put it to work in a real setting rather than a lab. They partnered with Crist Bros Orchards, a 500-acre commercial apple operation in Orange County, New York, where the robot was tested among actual trees. Field validation matters in agriculture because a tool that works on a test bench often stumbles on uneven ground, variable lighting, and the messy reality of a working orchard, and getting time in a commercial orchard is what separates a class project from a product.
The student who led the effort is Andrew James, a member of the class of 2026 studying agricultural sciences in CALS, who pulled together a group of agricultural specialists and engineers. The mix mattered. The agriculture students understood the problem growers actually face, while the engineers built the perception system and the arm, and the combination is part of why the judges singled the project out. The team developed its low-energy approach after studying existing electrical weeding technology and concluded it could do the job with far less power.
The win arrives against a backdrop of real pressure on farms. Labor shortages have made hand weeding expensive and unreliable, while tighter rules and grower concerns about herbicides have narrowed the chemical options, and weeds left unchecked steal water and nutrients and cut into yields. A robot that can move down a row, find the weeds that matter, and kill them with a jolt of electricity speaks directly to all three problems at once, which is why precision weeding has become one of the most watched corners of agricultural technology.
Rootline Robotics is already moving past the prize toward a real business. The grand prize was sponsored by Reservoir, which will host the team at its incubator in Sonoma, California, and the students are also advancing the technology through Cornell’s Center for Technology Licensing and the Rev: Ithaca Startup Works hardware accelerator, with continued field validation alongside grower partners planned through the summer. Turning a winning prototype into a machine that growers will buy and trust is a long road, with questions of cost, durability, and throughput still ahead, but the team has the early ingredients, a working robot, a commercial test site, and a clear problem to solve. For an idea built in four months by undergraduates, beating a field of seasoned competitors is a strong start.