FarmBot - Open-Source CNC Farming

Check out this time lapse of FarmBot watering the Swiss Chard that we planted last week!

You can see how this method of watering is much more efficient than soaking the entire bed with water:

We are often asked questions relating to FarmBot’s yield. “How big of a FarmBot do I need to grow all of my food?” is a common one. While we don’t yet have any empirical data to share with you, we have done an analysis answering this very question. Let’s dive in to see the results.

Gathering Data

We compiled a set of 33 common crops (seen in the table below) that could be compatible with FarmBot in the relative near future. You’ll notice that we did not include tall crops such as sunflowers and corn, and there are no fruit trees or berry bushes (with the exception of strawberries). Additionally, there are no grain crops included because it is unlikely that growing grains would be efficient with FarmBot hardware in comparison to larger scale specialized equipment.

After choosing the crops, we needed to find three pieces of data for each: average yield per harvest (kg/m^2/harvest), average days till harvest (days/harvest), and caloric density (calories/kg). We found this information from the USDA and a wide range of other sources online (see below for resources list).

Crunching Numbers

Right off the bat, we made an assumption that FarmBot can increase yield per harvest by about 12% by packing plants in a denser hexagonal packing structure instead of a traditional cubic packing structure. Using this new yield data and the days/harvest values, we calculated daily yield values for each crop in kg/m^2/day. Multiplying this figure with the caloric density provided a daily caloric yield for each crop in calories/m^2/day.

From here, we found out how many calories/day could be produced with a FarmBot Genesis (4.5 m^2 in size) and a FarmBot Genesis XL (18 m^2). We calculated these values twice: once by using an average caloric yield of all 33 crops to represent growing them all using an equal amount of area; and once by using an average of the best 10 performing crops as ranked by the calories/m^2/day benchmark. The results are in the table below:

As you know, 328 calories/day with a Genesis XL growing the best 10 performing crops is not very many calories. Most people eat at least 2,000 calories/day. By this analysis, one would need a huge FarmBot to grow all of their caloric needs. We calculated just how big in the table below, revealing a minimum size of 110 square meters (the size of a small house) in order to provide all 2,000 calories/day.

Cups, not Calories

But this isn’t a very useful calculation because most people don’t get 100% of their calories from vegetables. Not even close to it, actually. Most people also eat grains, dairy, meats, oils, fruits, etc – all of which are much more calorie dense than veggies. According to My Plate (previously the food pyramid), we’re supposed to eat about 3cups of veggies a day, and there is no mention of calories of veggies at all.

This means that the question we are asking and answering should instead be: “How big of a FarmBot would I need to grow the recommended cups/day of veggies I need?” Let’s see how this changes things.

If you chose to eat 3 cups of black beans every day, that would be over 1,500 calories, and require about 250 square meters to grow. This is a huge amount of area needed to satisfy our 3 cups goal. If you chose to eat 3 cups of arugula every day (a crop with similar calorie/kg value to black beans), that would be less than 20 calories, and only require about 3 square meters to grow. This is a much smaller amount of area needed because our new goal is cups, not calories. So while both crops perform similarly according to the calorie benchmark, they perform extremely different under the cups benchmark.

Promising Results

Looking back out our table of 33 crops, with new columns for cups/m^2/day, and FarmBot is looking quite a bit more feasible for satisfying our needs.

Using the average yield for our 33 crops, and an assumption that they will all be grown in equal quantities, one only needs about 7 square meters of space to grow the daily recommended veggies for one person. If one decides to grow only the top 10 performers (by the cups/m^2/day benchmark), then they will only need 3 square meters to grow all of their daily veggies.

This indicates that FarmBot Genesis could produce all of the veggies needed for one person every day. Meanwhile, a Genesis XL could produce enough veggies for a family of four every day. See the table below for the numbers:

We think that this data is quite promising for the FarmBot Project because it shows the capabilities of FarmBot in a clear and transparent light. Your FarmBot (unless it is very large) will not be able to produce all of your calories. However, it will be able to produce enough cups of vegetables to satisfy your daily needs.

What do you think of our analysis? We’d love to hear your feedback and ideas in the comments.

Resources used

Caloric density data from the USDA:
http://ndb.nal.usda.gov/ndb/foods

Average yield data from the following sources:
http://www.johnnyseeds.com/assets/information/vegetablecharts.pdf
Wolfram|Alpha Knowledgebase, 2015
http://www.gardensofeden.org/04%20Crop%20Yield%20Verification.htm
http://www.lsuagcenter.com/en/lawn_garden/home_gardening/vegetables/Expected+Vegetable+Garden+Yields.htm
http://aggie-horticulture.tamu.edu/vegetable/files/2011/10/swisschard.pdf
http://anrcatalog.ucanr.edu/pdf/7221.pdf
http://edgecombe.ces.ncsu.edu/files/library/33/production%20data%20table%20edit[1].pdf

Days to harvest data from the following sources:
http://www.ipm.iastate.edu/ipm/hortnews/2004/7-23-2004/vegguide.html
http://www.harvesttotable.com/2010/06/vegetable_harvest_times/

Recommended daily vegetable intake value from My Plate:
http://www.choosemyplate.gov/vegetables

FarmBot needs to be designed with manufacturability in mind – not just on an industrial manufacturing scale, but on the individual maker/hacker scale as well. After all, our goal is to empower people to build FarmBots! With this in mind, we have designed much of our hardware to be manufacturable using common tools and processes such as 3D printing, drilling, and sawing; as well as common materials including standardized hardware, aluminum extrusions, and aluminum connecting plates.

We’d like to share with you how we manufacture our 5mm thick aluminum connecting plates – an integral component of FarmBot that connects the extrusions, V-wheels, motors, and other components together throughout the hardware system. While the plates can be easily produced using a drill press and a hacksaw at home or in a MakerSpace, we’re going to share how we manufacture them in bulk with a high degree of accuracy.

The process we use is called water jet cutting, where a high pressure stream of water with an abrasive compound (often diamond dust) mixed into it is used to cut through a variety of materials. The machines are often very large and very expensive, though very powerful and versatile. Most machines can cut virtually any material including plastics, woods, aluminums, and steels. The process lends itself well to manufacturing our 5mm thick aluminum plates.

At FarmBot.io, we’ve been fortunate to work with a fantastic water jet cutting service provider named Big Blue Saw. Big Blue Saw is attractive for us to use because they are an online-first service that really caters to their customer’s needs. On their website you can upload a .DXF file of your parts, choose your material and quantity, and receive an instant quote for free. If the price looks right, you can place the order right then and there, and your parts will arrive at your doorstep within a week or two!

So far we’ve purchased about 12 FarmBot’s worth of aluminum plates over three different orders from Big Blue Saw. We’ve been very happy with their service, the fair price, and the quality of the components such that we’ll be continuing to go back to them for all of our water jet cutting needs.

If you’re planning on manufacturing your own FarmBot, we highly recommend downloading our .DXF file for all of FarmBot’s plates, and sending the job over to Big Blue Saw. There is also a basic finishing option for components, though we haven’t tried that out yet because we’re still just building prototpyes. However, once we begin selling FarmBot kits for others, we’ll be sure to polish up all of the plates for a professional, high-quality aesthetic.

We just wrapped up the documentation for FarmBot Genesis V0.8! In this version, we added bills of materials for each sub-assembly, more detailed and photo heavy assembly instructions for each sub-assembly, and details for setting up the software and electronics.

If you have any questions or issues, drop us a line in our support forum.

Two years ago I published a whitepaper describing an open-source technology called FarmBot and a vision for an Open Food Future. Since then I’ve been building two global communities to help me carry out this vision: the free-form FarmBot Project and the non-profit OpenFarm.

As the vision has developed, it has become important to make clear the differences between the entities. Sure, I’m the guy behind both The FarmBot Project and OpenFarm, but the two products, teams, legal entities, and finance books are very different and separate. OpenFarm is a non-profit and a database for how to grow food. The FarmBot Project is a free-form global community working to develop a common technology and sharing resources like the FarmBot Forum and Wiki.

However, while free knowledge for building things is important, sometimes it is not enough to get mass adoption of a new idea. Sometimes business can be a powerful agent of change that can make something more accessible. We see this a lot with the RepRap project and the hundreds if not thousands of 3D printer businesses that thrive around consumerizing and improving upon a core of open-source technology.

This is where today’s announcement comes in. Up to this point, we (me, Rick, and Tim) have been operating under The FarmBot Project as just everyday people working on FarmBot technology. This has worked out great for the last two years, though we’ve had larger ambitions to commercialize for some time now. Now we want to start a for-profit business producing and selling FarmBot hardware kits and software services. In doing this, we want to make clear the differences between our for-profit ventures and the free-form community behind the technology as a whole.

So today I’m excited to announce a third entity to the mix: FarmBot.io, our for-profit company that we’re forming to design and sell FarmBot hardware kits and software services. Though this is a private company, we hope to run it using open company practices and we’re committed to continuing to contribute our development back to the open source core of The FarmBot Project and community. We’ll also continue to maintain and steward The FarmBot Project’s community resources.

To make it all easy to remember and distinguish, we have two domain names that you should keep in mind:

• farmbot.cc is the community site for The FarmBot Project as a whole. This is where anyone can get involved and share ideas.
• farmbot.io is our business site where we design, sell, document, and provide support for our hardware kits and software services

We’re also starting new social media profiles for farmbot.io. Below is a breakdown of all the profiles:

• The FarmBot Project on Facebook
• The FarmBot Project on Twitter
• The FarmBot Project on Google+

• FarmBot.io on Facebook
• FarmBot.io on Twitter
• FarmBot.io on Google+
• FarmBot.io on AngelList

We hope this is clear now, and that it will only become more clear as time goes on. We think that as the project community grows, and as more for-profit businesses start popping up (which we encourage!) then the distinction will be more obvious.

Cheers to a prosperous future of FarmBot community and business!

Over the last month, I went on a roadtrip across the United States. It was mostly a vacation, though also an opportunity to hand deliver a V0.8 FarmBot to Rick in Chicago and film him talking about our software in preparation for our Kickstarter campaign coming up later this year!

The hardware installation process was pretty straightforward because I had originally had the FarmBot installed in my front yard, then broken it down ~20% into sub-assemblies for transportation, and then we simply put the sub-assemblies back together on Rick’s raised bed. The sub-assemblies were: Track 1, Track 2, Gantry Column 1, Gantry Column 2, Gantry Main Beam, and the Cross-Slide + Z-Axis combo. The hardware installation only took about an hour.

However, then we ran into some troubles. Rick built his raised bed about 200 feet from his house, so we had to string together a series of extension cords to provide the bot with power. Some of the cords were really old and caused the house circuit breakers to flip a few times, and even shock Rick once! But we eventually found better cords and got it powered up.

Then the device booted up and automatically connected to my phone’s wifi hotspot (because my home wifi network was not available anymore, and the bot had no way of knowing Rick’s home wifi network name and password). This was all going according to plan. Using my phone, I SSH’ed into the Raspberry Pi and edited the wpa_supplicant.conf file which controls which wifi networks the pi knows to connect to. I added in Rick’s home network and restarted the device. This is where stuff went downhill.

Upon restarting the device, it would not connect to either Rick’s home wifi, or my phone’s wifi. We tried moving Rick’s wifi router closer for a stronger signal, but to no avail. We then ended up bringing the Pi inside and double checking the wifi file using a linux computer. With everything looking good, the device would still not connect to any wifi networks. After several hours of troubleshooting, we never figured out what went wrong, and decided that the Pi will need to have a fresh OS installed. Oh well. This gives us reason to work on “FarmBot OS” – an image of a working OS with all of our software pre-installed on it so that users can flash it onto their Pi’s SD card and everything will just work.

In the meantime, Rick is pretty excited to finally have a full-size device of his own to test out software with!

We’ve just wrapped up FarmBot Genesis V0.7 documentation! This version (and past versions) are now hosted on our new documentation platform: readme.io. We’re really happy with how professional, useful, beautiful, and easy to follow these docs are. Since the last version of docs, we’ve added more assembly instructions, a more complete bill of materials, and new pages for software setup. Let us know what you think of the new docs!

Good documentation is critical for any open-source project. Be it hardware or software or both, good docs help others understand the design intention, how to navigate the project, and how to go about making modifications. Without good docs, the project might as well be closed source.

Up until today, we’ve been posting our docs in several places. Hardware docs were hosted here on go.farmbot.it, and also on the community wiki; while our software was docs were hosted on GitHub. The problem with hosting hardware docs on a wordpress site or on a wiki is that neither platform is particularly suited to making or presenting good documentation. This makes for some pretty ugly looking docs that are somewhat painful to create. Software docs on GitHub are great for stuff like: How to setup and run the software, but not so good at teaching software developers how to use an API, or the inner workings of a codebase, or how to go about modifying and extending the software. For a while, we’ve been needing a better solution for both hardware and software docs.

A few months ago we stumbled across a new documentation service called readme.io. Its a pretty new service aimed at creating beautiful docs and developer hubs, and we’re giving it a shot. So far, we’re very impressed with its ease of use and the beautiful docs it creates. What’s great about it is that it works well for both hardware and software, and it allows us to easily create new documentation versions based on old content. Plus, they are very supportive and generous in offering open-source projects free service!

So today we’re announcing two new documentation hubs:

• farmbot-genesis.readme.io is our new home for all documentation for FarmBot Genesis including change logs, design intention, assembly instructions, maintenance advice, and software setup
• farmbot-software.readme.io is our new home for documenting the web application and API, farmbot-serial, farmbot-arduino-firmware, and the farmbot-raspberry-pi-controller

Note: Both are still works in progress!