On the UC Santa Cruz farm, a small self-driving tractor navigates itself over a line of broccoli, avoiding the crops while removing the weeds that may interfere with their growth. This autonomous weeder is the work of a team of students co-advised by Professor of Electrical and Computer Engineering Dejan Milutinovic and Executive Director at the UCSC Center for Agroecology Darryl Wong, and has been recently recognized with the Small Farm prize at the inaugural Farm Robotics Challenge.
Members of the “Electrified Slugs” team Joshua Gamlen (computer engineering), Katherine Rogacheva (robotics engineering), Mauricio Chavez (agroecology), and Oliver Fuchs (robotics engineering) collaborated to create their submission for the Farm Robotics Challenge, a competition which presents students with real-life farm automation scenarios in order for them to gain practical experience with robotics in agriculture.
“This project is really exciting because it’s bridging what sometimes can feel like really distinct worlds,” Wong said. “It speaks to the strength of UC Santa Cruz and how it approaches Ag Tech, that we can have these conversations that seemingly come from different worlds, but do a really good job to inform and support each other.”
Nineteen student teams from 12 universities across the country participated in the challenge, and all teams had access to small small electric tractors from Farm-ng, a Watsonville-based robotics hardware and software company, upon which they had to build their solutions.
The Electrified Slug’s project focused on the development of the autonomous navigation software to allow the electric tractor to safely and efficiently weed plant lines on small, diversified organic farms, using the UCSC farm as a model for this problem. They chose this project focus after identifying labor costs and weeding as two of the biggest challenges facing local organic growers in California. Hand weeding and stooping can contribute to significant musculoskeletal issues for farmers and farm workers in organic systems.
To develop their solution, Milutinovic and his students used feedback control navigation, which is based on an original nonlinear mathematical model for relative motion between the tractor’s weeding tools and a plant line. They implemented the navigation on the Farm-ng tractor’s onboard computer and created a user interface which allowed the navigation to use visual information from the tractor-mounted camera. Users have the flexibility to mount the camera at a desired position to see the tractor tools and to identify plants without much calibration or intervention.
Importantly, this vision-based navigation can uniquely function across a wide range of crops as would be found on a diversified organic farm, a distinct difference from larger monocrop operations.
“At this point we are pushing the envelope of what is possible in food production,” Milutinovic said. “And it's not naive since we have to deal with uncertainties of operations on real farms. Robotics have decades of development, but we are just now in the position to get these robots outdoors and do some real stuff. That's really exciting.”
“The part of this project that feels most exciting is making sure we're approaching this nascent and potentially really disruptive form of innovation with a lens of inclusive and responsible innovation, to account for what we know are the harmful things that have happened historically during the process of these innovation cycles,” Wong said. “We're engaging in these conversations about why it is important that we understand how we frame innovation, and what we're trying to innovate toward — is it really just about efficiency or is it about people in the communities who are growing this food?”
The project has garnered enthusiasm for socially-impactful agricultural technology among students, both those on the team and those who work on the farm and have watched the tractor’s development. Several students from the team have accepted internships or full-time positions in related fields, and both advisors noted that students not involved in the project were curious to learn more about the tractor and the team’s work.
All teams in the competition, which included many of the traditional powerhouse agriculture schools, were judged on multiple criteria including design elegance and ease of use, interdisciplinary inclusion, and societal and economic impact. The competition was put on by AI Institute for Next Gen Food Systems (AIFS), UC Agriculture and Natural Resources VINE, Fresno-Merced Future of Food (F3) Innovation, and Farm-ng.
With their win of the Small Farm prize, the team received $5,000 and a trip to Salinas in September to present at FIRA USA, a networking and demo event for autonomous farming and agricultural robotics solutions.
This research has been supported by a $40,000 CITRIS Campus Seed Funding and additional private gift funding.
