TEDxUCLA 2014: Open 2.0

Open-source automated precision farming


About Rory

Rory Aronson is the creator of FarmBot, humanity’s open-source precision farming machine and software suite designed for small scale automated food production. He is also the creator of OpenFarm, a free and open database for farming and gardening knowledge. He resides in San Luis Obispo with 9 roommates, 2 dogs, and 6 chickens.


Hi everyone, my name is Rory, mechanical engineer, social entrepreneur.

Agriculture, it’s the world’s most important industry. And agriculture is going through some growing pains as our global society changes faster than it ever has before. The population is rising, the developing world is eating more meat. We’re physically running out of resources and space.

In a 2012 report by World Wildlife Fund, they state that humanity must now produce more food in the next 4 decades than in the last 8,000 years of agriculture combined. And we must do so sustainably. That’s a monumental challenge that we face.

And what I found is that in our race to adapt, in our race to feed the world, two major farming paradigms have come to dominate the landscape. And today I want to tell you about an idea for a third.

But first let’s start with the status quo. On the one hand we have the polycrop, a system where multiple types of plants are in the same area, mutually benefiting each other as well as the soil.

It’s a great system, here’s an example: my backyard garden. We have fruit trees, sunflowers, broccoli, kale, spinach, all working together; it’s an ecosystem at work, it’s very biologically efficient, and because of that we don’t have to add fertilizer or pesticides to make this function.

However, the polycrop is very labor intensive. It takes a lot of time to make this garden work. And on a commercial scale, less and less people want to do manual labor as a profession.

Now, on the other end of the spectrum, stemming from the Industrial, and Green Revolutions, is the monocrop. Here’s a shot from the Central Valley of California. It’s a paradigm that has reduced the ecosystem down to a single plant type, such that a machine, a tractor, can tend to all of these plants in the same fashion.

From a mechanical efficiency standpoint, an automation standpoint, this is great: Very few people growing tremendous amounts of food. However, this system, because it’s only a single plant, requires fertilizers and pesticides just to sustain itself. And these pesticides and fertilizers are very damaging to the environment, and also the food itself.

So let me tell you a little story. About three years ago, I took an agriculture class. I was hungry to learn more about this whole situation. One day, an industrial farmer came in, and he was so excited to tell us about his newest tractor, one that used a camera and a computer vision system to detect and destroy the weeds.

He went to the chalkboard and he drew us this simple diagram to demonstrate how it functioned. Those green dots are his lettuce plants in a row, and the red X’s are the weeds.

Now, to get rid of the weeds, the tractor would drive slowly down the row with a big hook tool, it’s pathway shown here in white, and this hook would physically disrupt the root systems of the weeds and bury the infant plants under the soil.

Now, here’s where it gets cool: When the camera and the computer vision system detected a lettuce plant, it would signal to the hook tool to skip a beat passing completely around the lettuce plant and keeping it intact. Pretty amazing technology.

Now, at the time, three years ago, this innovation cost half-a-million dollars. A huge sum of money. However, for this farmer, this was more economical, faster and more thorough at removing the weeds than hiring a dozen laborers to do this by hand season after season.

So looking at this from a backyard gardener’s standpoint, I thought, “Wow, that’s really cool, but where is the low-cost, small-scale version of any sort of agriculture technology that I could use in my backyard to help me grow more food more successfully?”

And from an engineer’s standpoint, I thought, “Wait a minute. Let’s step back. That’s actually kind of a duct taped on solution, an incremental change to a very old technology. The tractor hasn’t really changed that much in the last 100 years. Sure, they’ve become larger, but just now are we retrofitting these tractors to be more precise?”

We’re sort of duct taping in a precision computer system into a historically imprecise, free-driving, human-operated tractor. Why don’t we think about that and say, “Well, what if we built something new from the ground up?” At that moment, I had an idea to do just that.

There are plenty of machines out there that perform precision operations in, let’s say, an XYZ space. This CNC router, for example, cuts out wood shapes very precisely. So what if we took this concept and adapted it for growing plants? It might look something like this.

Meet FarmBot, this is an automated, precision farming machine. FarmBot plants seeds at very specific locations. Each plant has coordinates, and then FarmBot positions other tools very precisely in relationship to those plants in order to destroy the weeds, water the plants, and even sample the soil.

Before we had my backyard polycrop, biologically efficient but very labor intensive. We also had the monocrop of the Central Valley, automated but industrially harmful. With FarmBot, everything is automated, it’s all computer-controlled from the start.

And because we can plant multiple types of plants in the same area, it’s a polycrop that is automated, a third paradigm of farming, a hybrid of the other two combining the best of both.

In fact, by setting the tractor aside and reinventing, reimagining the backbone machine of food production, we open up doors to improve efficiency, and change the game of what is possible in agriculture.

With FarmBot, plants don’t have to be in a row; we could plant them in denser, more space-efficient arrangements. Soil compaction is nonexistent as the weight of the machine is placed completely on the tracks. Farming is done smarter with data from the weather report and sensors driving most of the operations.

By using low-cost electronics, and fabrication techniques made popular by the maker movement, FarmBot could be produced in a maker space, a FabLab, or even in your garage, such that it makes sense for you or I to have it in our backyards.

By programming FarmBot with a graphical web-based interface like a video game, anybody could be a farmer, no matter how little time, experience or even physical ability they have. Think Farmville, but in real life. (Laughter)

About eight months ago I wrote a paper describing this technology, the vision, potential risks — everything I’d ever thought about, I wanted to share it. And to do so, I published this paper freely on the Internet. I open-sourced the idea.

What’s mine is now yours too. It’s ours. Let’s collaborate, let’s build FarmBot together, because a challenge as large as reinventing food production needs a global team to pitch in and make it happen.

And within days of publishing this paper, engineers, software developers, farmers, gardeners, garage tinkerers began contacting me, sharing their ideas on the Wiki, the forum, in email. It was recognized that not only is FarmBot a cool and interesting idea, but that the larger technical and societal changes it could bring about are powerful and transformative.

At home, automated food production with complete control of the operation vested in the machine’s owner: Set it and forget it, FarmBot will email you when the tomatoes are ripe. And you can rest assured that those tomatoes are grown to your exact specifications: You know what went into them, you have control over the process.

Scalable, modular, low-cost hardware that is hackable for different applications. An open-source hardware, software and data ecosystem, based on sharing and collaboration, thereby encouraging innovation and the production of this technology. Perfectly optimized resource usage, spacing and timing for every single plant grown, and, of course, an automated polycrop, the third paradigm of farming.

Today, the FarmBot project team is over 25 people strong. We have six prototypes around the world. Here’s some electronics in Poland. This is a very small-scale FarmBot in Belgium, the first one to ever be watering.

Here’s the electronics that drive that machine. These guys are some mechanical engineering students in California. They’re working to develop a universal tool mount system so that FarmBot can change tools automatically.

Here’s a seed injector that we prototyped from a vacuum pump, and some 3D-printed parts. Here’s version-2 of the gantry system. A very small-scale version that I’m working on in my maker space.

What we’re building is an open and accessible technology to aid everyone to grow food, and to grow food for everyone. We’re exploring how a global team can come together around an open idea to help solve a big challenge.

We’re using the open-source model as a mechanism for rapid prototyping and the quick dissemination of new ideas and improvements to everyone. If the technology and the model prove viable, and we think it already is, then we may be uncovering a part of the solution to one of humanity’s most pressing contemporary challenges. Thank you. (Applause)