AI Meets Agriculture: Building Smarter, Greener Food Systems

In this episode of The Q&AI Podcast, Bob Friday sits down with Professor Yongsheng Chen from Georgia Tech to explore the transformative role of AI in agriculture. From early neural networks to cutting-edge urban farming systems, Professor Chen shares his journey from environmental engineering to building AI-powered agricultural ecosystems. Tune in as they discuss precision farming, hyperspectral imaging, decentralized food systems, and the future of urban agriculture powered by reclaimed resources and autonomous 

-----

Key points covered: 

AI’s evolution in agriculture: From early neural networks to modern machine learning applications in crop monitoring and yield prediction  

Precision farming with imaging and sensors: Using satellite, drone, and robotics data to optimize fertilizer and pesticide use 

Urban farming and decentralized agriculture: Repurposing urban spaces for sustainable food production using AI and reclaimed resources 

Georgia Tech’s pilot systems: Real-world implementation of controlled environment agriculture using dorm wastewater and AI-driven monitoring 

Networking’s role in smart agriculture: How connectivity and real-time data enable autonomous farming systems and decision-making

-----

Where to find Prof. Yongsheng Chen 

LinkedIn: – https://www.linkedin.com/in/yongsheng-chen-2823521b4/  

Where to find Bob Friday  

LinkedIn: – https://www.linkedin.com/in/bobfriday/ 

-----

Keywords

AI in agriculture, Urban farming, Decentralized agriculture, Controlled environment agriculture, Precision farming, Hyperspectral imaging, Machine learning in farming, Sustainable food systems, Agricultural robotics, Vertical farming, Reclaimed water irrigation, Smart agriculture, AI-powered crop monitoring, Networking in agriculture, Georgia Tech agriculture research 

-----

To stay updated on the latest episodes of The Q&AI Podcast and other exciting content, subscribe to our podcast on the following channels:

Apple Podcasts - https://podcasts.apple.com/us/podcast/the-q-ai-podcast/id1774055892

Spotify - https://open.spotify.com/show/0S1A318OkkstWZROYOn3dU?si=5d2347e0696640c2

YouTube - https://www.youtube.com/playlist?list=PLGvolzhkU_gTogP5IBMfwZ7glLp_Tqp-C

We hope you enjoyed this episode!

Chapters

• 0:04 Welcome and Introduction
• 0:58 Neural Networks in the 1990s
• 2:14 Transition to AI in Agriculture
• 5:04 Hyperspectral Imaging for Crop Monitoring
• 6:03 Urban Farming Initiative at Georgia Tech
• 9:34 Future of Urban Agriculture by 2040
• 12:34 Episode Wrap-Up and Farewell

Transcript

Bob: Welcome to another episode of Q&AI. Today we have the honor of Professor Yongsheng from Georgia Tech's Civil Environmental Engineering Department to join us, and today we're going to be talking about AI in agriculture and specifically about distributed decentralized agriculture. Welcome, professor. Maybe we can start with a little bit of introduction. Let the audience know who you are. 

Yongsheng: Thank you, bob, for having me on the show. My name is Yongsheng Chen from Georgia Tech. I'm the professor since 2009. I joined Georgia Tech. I'm also the director of New Center Nutrients, energy, water Center for Agriculture Technology. 

Bob: Yeah, I have to say, like you, I played around with neural networks in the 80s for my master's thesis back then. I'm just curious in the 1990s, how big were these neural networks you were using? I'm sure they weren't 500 billion weights. Do you remember how large these things were? 

Yongsheng: Actually, the computational resources was limited so we used a fortune to do our coding. So that is very, very challenging at that time and at that time it was a 386 type of computer. So, anyway, so that is the yeah, the challenges. But nowadays, you know, the computational power increased significantly, so it allows us to do everything we need to do with the large data sets. 

Bob: Yeah, so when I look at the history of these large neural networks and models and we went from kind of the research to chat GPT I'm kind of curious. You know, when you've watched the whole industry go from the 90s to where we are in the 2020s here, when did things really take off? When did these models get really large, when they go from kind of something we're playing with to something more interesting? 

Yongsheng: I think for this type of data-driven model we need to have a lot of data. At the same time, we need to make sure we are using the AI more efficiently to address the engineering related questions properly, not overly using the AI for different type of applications, for example for system optimization you don't have to use lots of data to generate a model. Actually, you can use that offline data to train your model and plug that model into online systems to further training the model. 

Bob: Well, it sounds like you've been playing around with these neural network models since the 90s, on the agricultural front. When did agriculture become relevant to your career? Did you leave school knowing that you're going to be working on AI in agriculture? 

Yongsheng: Yes, I think my transition from environmental engineering to AI-driven agriculture stemmed from a deep commitment to systems thinking. So environmental engineering told me that solving the real-world challenges demands interdisciplinary, system-level approaches. Agriculture quickly emerged as a critical domain.  

It contributes nearly 30% of global greenhouse gas emissions, with nitrous oxide from the fertilizers being a major driver. To address this, I began integrating tools like hyperspectral, multispectral images, 3d reconstruction, internet of things, sensors and machine learning into agricultural systems. It is not just about solving isolated problems. It is about connecting the dots across domains to design resilient, sustainable systems. 

Bob: You started to go from kind of the environmental component to more of a systems agricultural issue around and it looks like you're using satellite imagery. You're basically starting to understand how fertilizers are being used broadly across these big farms. 

Yongsheng: We have a project, actually funded by USDA, to actually use a different type of scales images satellite images, drone images and also local robotics images to come up with a model to predict the sugar bees' productivities in the Nebraska area. 

Bob: Okay, so really the goal of this is really try to minimize the amount of pesticides and fertilizer that's getting into our food supply. Was that kind of the vision around the effort food supply 

Yongsheng:  That is related to the hyperspectral imagines and the machine learning for the crops, so that we can use that as a tool to develop a long destructive high resolution monitoring across the large fields or within the controlled environments, supporting the precision agriculture by enabling timely interventions and optimized resources use. 

Bob: Now I have to say, when I met you at Georgia Tech, what got me excited. When I first met you, you were talking about decentralizing agriculture, urban farming. So, when did the urban farming start to become relevant to your work and your research?  

Yongsheng: That's almost more than 10 years ago. We have received USDA-funded project to use municipal wastewater as irrigation water, so at that time we started to work on that project. It's funded by USDA, since we are not an underground university at Georgia Tech, so we received five million dollars from USDA, which is one of the largest grant Georgia Tech has ever received since 1980. So, the overarching goal for since then we have formed our new center for agriculture technology.  

So, the overarching goal of this center is to integrate AI-powered urban resources recovery technologies with decentralized, controlled environmental agriculture. Autonomous vehicles and flying cars are on the horizon points to replace personal vehicles, public transportation and taxis. As these innovations reshape mobility, we will also liberate vast areas of urban land currently used for parking. This presents a unique opportunity to repurpose these spaces for decentralized urban waste into valuable resources. Coupling these systems with AI-powered controlled environment agriculture, we can grow food locally and sustainably, right where it is needed.  

Bob: Okay. So, the vision here, if I get this, this is vertical farming, urban farming. Your vision is hey, Waymo, Uber, we get into self-driving cars, all those parking lots we're going to repurpose as farms again. Now, is there any vertical farming here? Are we basically looking for flat spaces in the city to grow things? Or where do you envision all these fruits and vegetables being grown in the city of the future? 

Yongsheng: By integrating localized cultivation systems with circular resources recovery technologies. By leveraging innovative approaches such as vertical farms, rooftop greenhouses and AI-powered modular controlled environmental agriculture, food can be grown directly within the cities, dramatically reducing the transportation emissions, supply chain vulnerabilities and post-harvest losses. What sets these systems apart is their ability to utilize decentralized recovered resources, such as reclaimed water, nutrients from municipal wastewater and organic compost from with the food.  

Bob: So, Professor Young, you know looking at, you know the vision of having urban farming and actually transforming this into reality. You know I think you talked to me a little bit about you're actually doing your research projects. You actually have something up and running on the campus of Georgia Tech. How close are we to this becoming from, you know, going from vision to something that I can actually buy my food from an urban farm. 

Yongsheng: That's right. So actually, Georgia Tech has invested lots of money to build on-campus pilot systems to treat dorm wastewater as resources to grow produce within the controlled environmental vertical farm. So, it is supported by a part of the USDA project. Georgia Tech come up with $1.5 million to build up the pilot site and also renovated our labs to build four controlled environmental agriculture chambers to allow us to grow produce under controlled environments. The AI related technologies to link that with the food yield prediction and nutritious prediction. 

Bob: And I would say, you know, I have to ask you know Juniper, big networking company, big AI company, you know how relevant is networking to solving this urban food farming problem? 

Yongsheng: So, I think network is very, very important. Without having that, it's very hard for us to make real-time decisions, so that is very, very important for urban agriculture. Meanwhile, the next generation robotics are moving beyond automation to autonomy, capable of adapting to changing the environments and executing the labour-intensive activities like precision planting, targeted pesticide application and selective harvesting, with unparalleled accuracy and consistency. 

Bob: Now, maybe to wrap it up here, professor Yongsheng you know, self-driving Waymo, uber taxis, those are real, those are here today. You know, maybe for the audience here, how soon am I going to be buying, going to the grocery store and be buying my fruits and vegetables from the local urban farm? You know, is that in my lifetime or is that for my kids? 

Yongsheng: I think by 2040. So once the disruptive technologies I mentioned atomic vehicles and also the flying cars and public transportations, personal cars, so that's where make the urban mobility totally changed okay. 

Bob: Well, I have to say that may be at the end of my lifetime, but it'll be definitely around my kids, but for us, or the only, I want to thank you for joining AI in Agriculture. I'm glad to hear that it's coming our way. I want to thank the audience for joining us for another episode of Q&AI and look forward to seeing you on the next one.