One of the thrust areas within the Minnesota Robotics Institute focuses on advancements in agriculture and environmental monitoring. Your lab’s recent work is very much centered in this area—what are a few motivations for exploring projects in these fields?
The state of Minnesota has strong roots in agriculture. Many industry leaders are stationed within and around the Twin Cities. Robotics can have a strong positive impact in these areas in multiple ways: we can create efficient and cleaner ways and to survey land to monitor plant life, soil and water. Beyond surveying, robots can perform farm tasks such as picking and weeding in environmentally friendly ways.
On the other hand, agriculture is an exciting frontier for robotics research in our efforts to move robots out of structured environments, such as factories.
Could you outline some of your team’s current projects?
We have multiple projects on surveying. The most prominent one is focused on pre-harvest yield estimation for specialty crops such as apples. From a sequence of images, we can estimate parameters such as the number of fruit, their size as well as tree volume and trunk diameter. We are also exploring projects in robotic picking and autonomous weeding in pastures and row crops. Finally we have multiple projects on using drones to survey farms and animal habitats.
I’ll start with that first project you mentioned—could you talk more about the automation process for these specialty crop fields?
Yes, sure. Our project has a few components. The main part is to detect apples and track them across images. This is a difficult problem because the fruit might be occluded by branches or other fruit. Or light conditions can change drastically. We developed a very robust method for counting and measuring apples and are now looking into ways to commercialize this technology. Other components of this system include autonomous UAV flight through orchard rows (which can be used to collect imagery) and using near infrared measurements to assess fruit quality. This project is in collaboration with Horticultural Sciences professors Cindy Tong and Jim Luby and twas supported by a MnDrive seed grant and evolved into a USDA project.
It’s clear that this and the other aforementioned projects fit well into the needs of Minnesota agriculture. Of course, this type of crop production is prevalent throughout the rest of the United States and beyond. Nevertheless, how important is the state of Minnesota and the associated industries in propelling and facilitating robotics research? And, more specifically, your team’s research?
The State of Minnesota, since 1988 has had a dedicated fund (Environmental and Natural Resources Trust Fund, or ENRTF) that is used to fund projects that protect, conserve, preserve, and enhance Minnesota's various natural resources, such as our effort to develop an automated mowing system. Being so intertwined within the state’s large agricultural industry and given the plethora of large companies looking to invest in robotics and automation, we have access to so many opportunities that would not be present in other states. Coupled with the effort made by the Legislative-Citizen Commission on Minnesota Resources’, who oversees the ENRTF, the state is the perfect place to be for this type of robotics research.
Could you elaborate on the automated mowing system your team is working on?
This is a collaboration led by our colleagues at the West Central Research and Outreach center at UMN Morris. The focus is on automated weeding processes in crop fields and improve forage quality in pastures. The system will be able to replace the weeding process that many farm currently use—reducing the amount of herbicides that can enter the water supply or affect plant life, replacing carbon emissions produced by machinery with locally produced energy, and introducing time-saving equipment to farmers. For pastures, Toro is also a partner. They are modifying one of their platforms. We are building the autonomy stack.
And, I believe you are also working on an aerial drone system?
We have multiple projects in this domain. An interesting project led by our collaborators at UC Boulder involves monitoring complex atmospheric phenomena. We are combining autonomous airborne sensors with environmental models to enable the collection of information essential for examining the fundamental behavior of atmospheric phenomena. The motivating application for this work is improvement in the accuracy and lead-time of tornado warning, which are also common in the state of Minnesota.
You keep mentioning your team and lab, can you tell us about your lab and your students?
None of this would have happened without the hardwork and dedication of my students. Our team currently consists of twelve talented students and postdocs. Our Ph.D. alumni have gone on to occupy positions at the academia in the US and Europe, NASA Jet Propulsion Labs, and Google. Dedicated undergrads working in our lab are now PhD students at UMN or other top schools. I cannot do justice to highlighting their accomplishments. Please visit my groups webpage, where you will be able to navigate to their individual web pages and read more. Thank you!
You can visit Dr. Isler’s website here and read more about MnRI’s educational and outreach opportunities here.
