Farming has been undergoing quite a revolution in the past few years. Since World War 2, most industrial farming has relied on synthetic fertilizer, large machinery, and huge farms with single crops. Now there is a growing number of successful farmers bucking that trend with small farms growing many crops and using natural methods of fertilizing that don’t require as much industry. Of course even with these types of farms, some machinery is still nice to have, so this farmer has been developing an open-source automated farming robot.
The robot is known as Acorn and is the project of [taylor] who farms in California. The platform is powered by an 800 watt solar array feeding a set of supercapacitors for energy storage. It uses mountain bike wheels and tires fitted with electric hub motors which give it four wheel drive and four wheel steering to make it capable even in muddy fields. The farming tools, as well as any computer vision and automation hardware, can be housed under the solar panels. This prototype uses an Nvidia Jetson module to handle the heavy lifting of machine learning and automation, with a Raspberry Pi to handle the basic operation of the robot, and can navigate itself around a farm using highly precise GPS units.
While the robot’s development is currently ongoing, [taylor] hopes to develop a community that will build their own versions and help develop the platform. Farming improvements like this are certainly needed as more and more farmers shift from unsustainable monocultures to more ecologically friendly methods involving multiple simultaneous crops, carbon sequestration, and off-season cover crops. It’s certainly a long row to hoe but plenty of people are already plowing ahead.
Thanks to [rvense] for the tip!
38 thoughts on “Automate The Farm With Acorn”
Ok, now I’m curious, I’ve noticed a lot of farm robots have that similar form-factor, why is that? I understand that the wheels are like that because of the terrain it has to work on, but the chassis of the robot is what I don’t get. Is it to make it as modular as possible?
I imagine that one wants a design that’s easy to build and easy to repair, especially on a farm. Plus making things easy for installation of panels, motors, and batteries, among other things. K.I.S.S in other words.
That form factor allows the wheels to travel between the rows of plants while the platform rides above the row of plants.
This is why.
This is the way.
Neat, but what does the machine actually do? Does it apply pesticides, plant anything, or harvest anything? Does it just show you where the weeds are so you can go out there and pull them? Maybe add a remotely operated arm for pulling weeds?
I never seen a pulling weed arm machine, I mean this is the proper way to remove bad weeds. Cutting them is pretty pointless.
I (recently) saw a farming robot that automatically identified weeds and used a laser to burn the leaves, the suggestion was that this robot going over the same ground at least daily would be able to prevent any re-growth with successive lasering.
Common solutions are to whack them in the ground, burn them with lasers, burn them with boiling water or to pluck them from the ground. The plucking solution reduces the max speed of the robot. The goal is not to ensure that the weed doesn’t grow back, but to simply fight a war of attrition. Removing the leaf often enough ensures it never grows too big and becomes a competitor to the crop.
Just cutting off the leaves decreases weed competition with the garden plant. While not a perfect or permanent solution, It’s not bad. Plus it leaves organic matter in the soil for stability and eventual incorporation.
From the linked article it’s doesn’t do any farming yet. It’s meant as a common platform for experimentation.
Not unlike the harvester platform I was writing code for…
This was a behemoth, powered from 4 truck batteries in series… that had two 20kW 48V hub motors in diagonally opposite wheels (each wheel was about 1.2m diameter), and all four wheels were intended to be independently steerable.
The driver control console took input from various pots and a rotary encoder to measure the driver’s intended course and speed, then its on-board processor (originally a MSP430F135, later MSP430F143, eventually it moved to a LM3S6965EVB) would compute the necessary rotation speed and angle of each wheel and send that information out over CANBus.
What was it intended to harvest?
I worked on wheat seeding robots waaay back in ’85. I’m really excited to see armies of solar powered cultivation/ seeding, weeding and harvesting agricultural robots. All the technology already exists, and the requisite engineering really doesn’t seem all that difficult.
Wow thank you so much for the coverage again on my robot! Also none of this would be possible without the support of Daniel Theobald, who owns the farm and has collaborated with me on this design. He originally devised the concept and I have been designing it for him for the last two and a half years.
Please sign up to our community site and join the discussion (link in the article). Thanks Hackaday!
My subconcious is adamant that a robot named Acorn must have a 6502 or Arm processor hiding in it somewhere…
and be running the kernel, of course.
Are any components supplied by Seeed Studio?
Needs more tiedye! Unleash the power of Mintberry CRUNCH! https://southpark.fandom.com/wiki/Mintberry_Crunch
This! I didn’t even see a robot in that picture
I learned long ago that fashion is not a high priority in farm work.
Is the robot actually farming anything ? It does not show there is even an attempt of farming / doing anything useful
Why all “farming robots” here are just roaming the fields instead of just focusing / tackling on the much tricky “farming” task ? putting the robot later on a kind of buggy like this one is a secondary task, derived from the first function of the robot!
You just call it “field-roaming robot that might one day do anything else”
You have a valid point but let’s not discount the large amount of effort and technology that went into make a robotic platform that drives itself around the fields entirely under solar power, that’s a pretty important foundation to build the next part on.
I too want to see robots pulling out weeds but there’s no denying that what has been achieved here is already quite significant!
The weed problem does not necessarily require a mechanical answer (zapping, pulling one at a time). Cover crops are already a good option in some cases. True biodegradable mulches and eco herbicides (if possible) are probably the way to go. Agro-robots are best suited for preparing the land, planting and harvesting. A large part of that tech already exists. In theory when a machine plant something with a precise distance, is not mandatory to have a complex computer vision guided robot arm to pick a plant or hoe it.
The correct way to automate farming is to bring it indoors – hydroponics now, direct tissue culture (ie just grow the corn kernels, not the whole plant) in future. Then those fields can get back to their natural state, which looks like woodland in this case.
bringing farming inside requires you to build a building right?where do you think all those buildings are going to go into those fields.
Floating on the ocean.
Because dolphins, fish and corals just hate sunlight, right?
In the sky!
If it’s good enough for Dorias Quanderhorn…
Indoor farming is extremely energy (fossil fuel in most of the world) intensive. You get free sunlight outside.
Most hydroponic nutrients are fossil fuel derrived. Compared to some practices it does conserve water so it makes sense in arid regions.
Unless you live at higher latitudes outdoor farming is superior in every way. Responsible practices like poly culture and low / no till methods improve the land while you farm.
Energy will soon be sustainable and cheap.
We should give the land back to Nature.
Sustainable agriculture is not the same as industrial agriculture. The first is able to increase biodiversity, sequester water and create soil much faster than Nature (btw man is also part of nature). The road to hell is paved with good intentions. It is not difficult to imagine a dystopian future where corporations control the world production of synthetic food and criminalize farming, ending the food independence of people. At the same time they buy thousands of acres to privatize, and give them back to nature…and demand payment from governments for the service (interestingly some millionaires have started making this type of investment recently – Tax free foundations). A Soylent Green for the people and grass beef for the elite, kind of future.
Ha, because agriculture now is just a collection of small community-minded artisans. No shady global monopolists clearcutting the Amazon, or anything like that, cough, cargill, cough.
Generating power, distrubutung it, and converting it into light will never be as efficient as using sunlight.
In most cases, using a few states worth of land for agriculture will have less impact than indoor high intensity farming using artificial light.
Great open project, impressive prototype platform and a interesting introductory video.
Seeing stuff like this being designed for operation in the wild is inspiring. Rough terrain, rain, mud!
My mobile RPi/JetsonNano platfom balks at thick pile carpet, overheats when it gets too dusty and died when I splashed a tiny bit of beer on it.
It’s capable doing everything, so it doesn’t do anything.
Maybe for planting seeds, but not even for watering, or spraying pesticides, but not in a bio farm.
I don’t buy the solar, as a farmer I’m happy to charge it overnight.
Also it would be a lot easier with line following, bury a wire, no need for expensive precision GPS.
Well, (Devil’s Advocate role) wires corrode, or get chewed through by gophers.
Energy will soon be sustainable and cheap.
We should give the land back to Nature.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)