Humanity's open-source automated precision farming machine.

Published by Rory Landon Aronson on September 19, 2013Email ­ rory@farmbot.itPhone ­ (678) 321 7679

Go.FarmBot.It

1

1. AbstractThe world’s population is growing and with that growth we must produce more food. Due to the industrial and petrochemical revolutions, the agriculture industry has kept up in food production, but only by compromising the soil, the environment, our health, and the food production system itself. The increased production has largely come from incremental changes in technology and economies of scale, but that trend is reaching a plateau. Conventional agriculture methods are unsustainable and a paradigm shift is needed.

FarmBot is an open­source and scalable automated precision farming machine and software package designed from the ground up with today’s technologies. Similar to today’s 3D printers and CNC milling machines, FarmBot hardware employs linear guides in the X, Y, and Z directions that allow for tooling such as plows, seed injectors, watering nozzles, and sensors, to be precisely positioned and used on the plants and soil. The entire system is numerically controlled and thus fully automated from the sowing of seeds to harvest. The hardware is designed to be simple, scalable, and hackable. Using the open­source web­based software package, the user can graphically design their farm to their desired specifications and upload numerical control code to the hardware. Other features of the software include storing and manipulating data maps, a decision support system to facilitate data driven farm design, access to an open data repository, and enterprise class analytics.

The following is a sample of intrinsic advantages of FarmBot that make it a superior system to conventional methods and technologies.

Ability to plant polycrops in a machine efficient manner Ability to optimize operations such as watering, spraying, and seed spacing Full automation and 24/7 possible operation Virtually unlimited farm design possibilities Incorporates “Big Data” acquisition and analysis for data­driven decision making and “Smart Farming” Ability to plant in the most space efficient layouts Scalable from a backyard system to an industrial operation Allows for the democratization and decentralization of food production Free and open­source, fully documented, hackable, and accessible

The vision of this project is to create an open and accessible technology aiding everyone to grow food and to grow food for everyone. The mission is to grow a community that produces free and open­source hardware plans, software, data, and documentation enabling everyone to build and operate a farming machine.

This white paper covers the FarmBot technology and vision as well as details of FarmBot Genesis, the first FarmBot currently in development with plans for a crowdfunded launch in 2014. I am seeking more people to work on the project and for investment to help cover prototyping and other development cost. Please send feedback and inquiries to rory@farmbot.it, call (678) 321 7679, and Go.FarmBot.It!

2

2. Table of Contents1. Abstract...........................................................................................................................................................................................................................22. Table of Contents............................................................................................................................................................................................................33. Background.....................................................................................................................................................................................................................64. FarmBot..........................................................................................................................................................................................................................7

4.1. Precision..........................................................................................................................................................................................................84.2. Combining Polycrop and Monocrop Efficiencies.............................................................................................................................................9

4.2.1. Monocrops........................................................................................................................................................................................94.2.2. Polycrops.........................................................................................................................................................................................10

4.3. Data­Driven Agriculture..................................................................................................................................................................................114.4. Automation......................................................................................................................................................................................................114.5. Scalability........................................................................................................................................................................................................114.6. Increased Space Efficiency............................................................................................................................................................................114.7. Eliminated Soil Compaction...........................................................................................................................................................................124.8. Continuous Land Use.....................................................................................................................................................................................124.9. Variable Terraforming.....................................................................................................................................................................................124.10. Open­Source................................................................................................................................................................................................13

5. Hardware.......................................................................................................................................................................................................................145.1. Tracks.............................................................................................................................................................................................................15

5.1.1. Geometry.........................................................................................................................................................................................165.1.2. Scalability.........................................................................................................................................................................................165.1.3. Cost..................................................................................................................................................................................................16

5.2. Gantry.............................................................................................................................................................................................................175.2.1. Geometry.........................................................................................................................................................................................175.2.2. Drive System...................................................................................................................................................................................185.2.3. Scalability.........................................................................................................................................................................................18

5.3. Cross­Slide.....................................................................................................................................................................................................185.3.1. Geometry.........................................................................................................................................................................................195.3.2. Drive System...................................................................................................................................................................................195.3.3. Scalability.........................................................................................................................................................................................19

5.4. Tool Mounts....................................................................................................................................................................................................195.4.1. Geometry.........................................................................................................................................................................................195.4.2. Drive System...................................................................................................................................................................................20

3

5.4.3. Scalability.........................................................................................................................................................................................205.5. Tools...............................................................................................................................................................................................................215.6. Electronics......................................................................................................................................................................................................225.7. Sensors...........................................................................................................................................................................................................22

6. Software.........................................................................................................................................................................................................................236.1. Web Frontend.................................................................................................................................................................................................23

6.1.1. User Interface...................................................................................................................................................................................236.1.2. Dashboard Tab................................................................................................................................................................................236.1.3. Farm Tab.........................................................................................................................................................................................246.1.4. Data Tab..........................................................................................................................................................................................286.1.5. Manual Control.................................................................................................................................................................................296.1.6. User Experience..............................................................................................................................................................................29

6.2. Backend..........................................................................................................................................................................................................296.2.1. User Profiles....................................................................................................................................................................................296.2.2. Farm Profiles...................................................................................................................................................................................296.2.3. Equipment Profiles...........................................................................................................................................................................296.2.4. Decision Support System................................................................................................................................................................29

6.3. Microcontroller Software.................................................................................................................................................................................316.4. Data Sharing and Open Data Repositories....................................................................................................................................................316.5. Mobile Applications.........................................................................................................................................................................................316.6. Open­Source and an Open Platform..............................................................................................................................................................31

7. Data...............................................................................................................................................................................................................................327.1. Plant Data........................................................................................................................................................................................................327.2. Soil Data..........................................................................................................................................................................................................327.3. Companion Plant Data....................................................................................................................................................................................337.4. Time and Location Data..................................................................................................................................................................................337.5. Weather Data..................................................................................................................................................................................................337.6. Topography.....................................................................................................................................................................................................347.7. Past and Future Data......................................................................................................................................................................................347.8. Manual Input Data...........................................................................................................................................................................................347.9. Downloadable and Auto­generated Farms.....................................................................................................................................................34

8. FarmBot Genesis...........................................................................................................................................................................................................358.1. Genesis Tracks...............................................................................................................................................................................................368.2. Genesis X­Direction Drive System.................................................................................................................................................................378.3. Genesis Gantry...............................................................................................................................................................................................378.4. Genesis Y­Direction Drive System.................................................................................................................................................................39

4

8.5. Genesis Cross­Slide.......................................................................................................................................................................................408.6. Genesis Tool Mount and Z­Direction Drive System........................................................................................................................................418.7. Genesis Tools.................................................................................................................................................................................................42

8.7.1. Genesis Seed Injector and Seed Bay..............................................................................................................................................428.7.2. Genesis Watering Nozzle................................................................................................................................................................43

8.8. Genesis Electronics........................................................................................................................................................................................448.8.1. Microcontroller..................................................................................................................................................................................448.8.2. Stepper Motors.................................................................................................................................................................................44

8.9. Cost.................................................................................................................................................................................................................458.10. Crowdfunded Launch....................................................................................................................................................................................478.11. Timeline........................................................................................................................................................................................................47

9. Potential Impacts, Ramifications, and Concerns of FarmBot.......................................................................................................................................489.1. Increased Production Efficiency.....................................................................................................................................................................489.2. Sustainability...................................................................................................................................................................................................489.3. Democratization of Food Production..............................................................................................................................................................489.4. Decentralization and Localization of Food Production...................................................................................................................................489.5. Elimination of the Farmer................................................................................................................................................................................489.6. Increased or Decreased Separation of People and Farming.........................................................................................................................499.7. Greater Dependence on Machines and Computers.......................................................................................................................................499.8. Loss of Individual and Generational Knowledge and the Gain of Universal and Accessible Knowledge.......................................................499.9. Lock­in to an Inferior System..........................................................................................................................................................................509.10. Hacking.........................................................................................................................................................................................................509.11. Failure of Supporting Infrastructure...............................................................................................................................................................50

10. Defining Short Term Success and Next Steps..........................................................................................................................................................,.5110.1. List of Action Items........................................................................................................................................................................................51

11. Conclusions.................................................................................................................................................................................................................53

5

3. BackgroundThe world’s population is growing and with that growth we must produce more food. Due to the industrial and petrochemical revolutions, the agriculture industry has kept up in food production, but only by compromising the soil, the environment, our health, and the food production system itself. The increased production has largely come from incremental changes in technology and economies of scale, but that trend is reaching a plateau. Conventional agriculture methods are unsustainable and a paradigm shift is needed.

The FarmBot project started my 3rd year of college (2011) while at Cal Poly in San Luis Obispo. I was studying Mechanical Engineering but decided to take an Organic Agriculture class to help fuel my interest in farming, gardening, food, and the environment.

One day, a guest speaker and local farmer came and spoke to the class. He was an elderly man but very up­to­date on his agriculture technology. He drew for us on the board a spiral looking line and turned around, very excited and pleased. He described to us how his newest tractor, equipped with a camera and a computer, could tell the difference between weeds and lettuce and selectively destroy the weeds without damaging the lettuce! The tractor used a tilling tool, similar to a pirate's hook, to churn up the soil in a spiraling fashion as the tractor drove forward, turning over all weeds in the path. When the camera and computer detected a head of lettuce in the path, the pirate hook would "skip a beat" and pass around the lettuce, keeping it intact. The system cost a good chunk of change, but with the long term vision in mind, would save money by reducing labor cost.

While staring at the chalkboard I couldn't help thinking: isn't there a better way of knowing where you planted your lettuce? In some sort of stroke of genius, I realized I could accomplish the very same task in a much more simple and elegant way. If I put the tractor, or more specifically the tractor tooling, on fixed tracks, I could know exactly where the tooling was located in relation to the ground and the plants, much like a 3D printer or CNC milling machine knows exactly where the tool head is in relation to the environment and surrounding objects.

Over the remaining two and a half years in school, I completed a bunch of research, sketched designs, completed some preliminary CAD modeling, built a visual prototype, and thought a lot about the project. The benefits of the system kept unfolding and it seemed to get more and more promising as the ideas brewed.

Now, the Summer after graduation, I have found a new vision to move the project forward and I am excited to dedicate a lot of time to it. I am continuing development on paper, in research, in writing, and with CAD. I am hoping to have a rudimentary working prototype by the end of 2013 and a crowdfunded launch in 2014. I am actively looking for more people to work on the project and for investment to help cover prototyping and other development cost. If you are interested, please email rory@farmbot.it, call (678) 321 7679, visit Go.FarmBot.It, and keep on reading!

Please note that this is a preliminary report of what I have developed and thought about thus far. This paper is largely speculative, I have limited experience with agriculture, and there may be many errors and unbacked claims.

6

4. FarmBotFarmBot is an open­source and scalable automated precision farming machine and software package designed from the ground up with today’s technologies. Similar to today’s 3D printers and CNC milling machines, FarmBot hardware (shown in FIgure 4.1) employs linear guides in the X, Y, and Z directions that allow for tooling such as plows, seed injectors, watering nozzles, and sensors, to be precisely positioned and used on the plants and soil. The entire system is numerically controlled and thus fully automated from the sowing of seeds to harvest. The hardware is designed to be scalable, simple, and hackable. Using the open­source web­based software package, the farmer can graphically design their farm to their desired specifications and upload numerical control code to the hardware. Other software features include storing and manipulating data maps, a decision support system to facilitate data­driven design, access to open data repositories, and enterprise class analytics. FarmBot has several distinct advantages over today’s methods and technologies that will be explained in sections 4.1 through 4.8 and the rest of the paper.

Figure 4.1. FarmBot high level hardware overview and coordinate system.

7

4.1. PrecisionAgriculture equipment and technology has been steadily becoming more accurate and precise with the advent of GPS and short range radio locationing, tractor autopilot, computer vision, data mapping tools, and variable rate tooling. These technologies are incrementally augmenting the abilities of free­driving tractors to allow some of the many benefits of precision agriculture to be realized.

Rather than making incremental changes to existing equipment, FarmBot takes a new approach at precision agriculture, tearing down everything from the past and starting from the ground up. By simply placing the tooling equipment on a set of tracks, rather than a free­driving tractor, the system has the ability to be extremely precise and reposition tooling in exact locations repeatedly over time. This is done with similar technology that has been around for decades in printers, manufacturing equipment, and more recently 3D printers and CNC milling machines.

This new method of precision agriculture has the potential to be as accurate as human labor with the automation and cost effectiveness of a machine. Precision can increase efficiency in the agriculture system in many ways including the following.

Weeds can be eliminated without damaging desired plants through selective burning, spraying of pesticides, or tilling. A mock up top­view of selective weeding is shown in Figure 4.1.1, where FarmBot can actively avoid desired plants when performing destructive operations.

Any type of plant packing structure can be created and managed including traditional cubic packing, hexagonal packing, and custom irregular structures. See section 4.5 for more details.

Each plant can be watered, fertilized, and sprayed individually and precisely with an optimized regimen that changes throughout the plant’s life cycle. Plants can be watered at the stalk, in a circle of a certain radius; sprayed everywhere, or just in problem areas, etc.

Plant life cycles do not have to start and end at the same time. Instead, any open space can be immediately replanted. Precision data mapping will allow the best plants to be grown based on spatial and temporal conditions.

Figure 4.1.1. By knowing where desired plants are, FarmBot can remove everything else.

8

4.2. Combining Polycrop and Monocrop EfficienciesFood production is all about efficiency. How might we produce more food with less water, fertilizer, labor, etc, without compromising sustainability? There exist at least two types of efficiencies in food production: biological and machine. The efficiencies have always been more or less exclusive, with biological efficiency stemming from the more natural polycropping system and machine efficiency from the machine intensive monocropping system. If we look at both the monocrop and polycrop, we can see where each one excels in efficiency, where each one lacks, and then see how FarmBot combines the best of both worlds.

4.2.1. MonocropsThe monocrop benefits from superior machine efficiency. The monocrop system shown in Figure 4.2.1.1 has reduced the farm ecosystem down to one plant species in order for today’s tractors and tooling to perform operations easily, reliably, quickly, with minimal human labor, and at minimal cost. This system is very conducive to scaling up, which is why we usually see the monocrop system implemented on very large farms with large tractors. However, the monocrop has perhaps zero biological efficiency, requiring many inputs to continue functioning. Because the ecosystem is so simple, it is unstable, unsustainable, and vulnerable to attack. Monocrops require more fertilizers, pesticides, energy, and water than any other farming system ever invented and it is still a struggle if not impossible to avoid depleting the topsoil, polluting the groundwater, and defending against insects and massive crop failure. Moreover, it can be argued that the monocrop produces less than the highest quality food.

Figure 4.2.1.1. Monocrops, require the most inputs of any farming system but are very machine efficient.

9

4.2.2. PolycropsThe polycrop benefits from superior biological efficiency. The polycrop shown in Figure 4.2.2.1 increases diversity within the ecosystem in order to be closer to a naturally occurring system, thereby reducing the number of inputs needed. Less fertilizer is needed with proper crop rotation and less pesticides and water are needed with intercropping. The plants work together to form synergistic relationships making the system more stable, resilient, and sustainable. These biological efficiencies come at a cost though. No traditional farming equipment exists that can perform operations on a wide range of plants at the same time, so machine efficiency is sacrificed and more human labor is needed to tend the crops.

Figure 4.2.2.1. Polycrops, are very biologically efficient but require much more labor due to a lack of large scale equipment for managing them.

FarmBot attempts to combine the efficiencies of both systems into one. It does so by being the first, scalable farming equipment that can perform operations on a polycrop. FarmBot incorporates the machine efficiency, scalability, and the minimal use of labor that traditional equipment takes advantage of while managing a polycrop with great biological efficiency. FarmBot’s abilities, combined with the knowledge of how to properly space plants, create synergies, utilize beneficial insects, and match nutrient users and givers, could enable the creation and management of abundant, resilient, sustainable, and efficient farms with minimal human labor.

10

4.3. Data­Driven Agriculture“Big Data” has been a hot topic recently and is transforming many industries. Today, agriculture data is costly to gather and the systems to make decisions based on this data are fragmented, proprietary, and not as powerful as they could be. FarmBot aims to make big data acquisition and analysis a more accessible and standard practice in agriculture, allowing all farmers to make smarter, data­driven decisions. FarmBot hardware and software will enable the farmer to routinely gather many types of data (see section 7) in a cost effective manner, map that data (see section 6.1.4), and allow the farmer and the decision support system to make optimized decisions for the layout and operation of the farm as described in section 6.2.4.

4.4. AutomationFarmBot will eventually become a completely automated system from the point of adding bulk inputs such as seeds and water, to removing bulk outputs like tomatoes. FarmBot aims to eliminate the need for human labor to drive tractors, pull weeds, harvest, and complete other operations. As the software and data analysis improves, the job of the farmer to create a farm layout and manage the operation of the FarmBot will also be eliminated in favor of downloadable and automatically generated farm layouts. By automating more of the processes, efficiency will be maximized through constant monitoring, optimized decision making, the minimization of waste and inputs, and the reduced need for human labor.

4.5. ScalabilityFarmBot is designed with scalability in mind. The hardware design intention allows scaling from a small garden sized machine all the way up to an industrial farming operation. The same software will be used in all applications with potential basic, intermediate, and advanced levels of control depending on the user’s experience. Because the system functions the same on every scale, FarmBot could disrupt the economies of scale of big agriculture, making smaller scale, more local farming more efficient for distribution channels and also increase resilience against severe weather conditions.

4.6. Increased Space EfficiencyFarmBot enables planting in a more space efficient packing structure, or layout, of plants that minimizes the space between them. Inspired by the hexagonal close packing of atoms, the most space efficient atomic structure, FarmBot allows for the hexagonal close packing of plants shown in Figure 4.6.1. This layout increases the amount of planted area by over 12% compared with the traditional cubic layout shown in Figure 4.6.2. This means one could grow 12% more food on the same area of land without decreasing the space each plant needs.

Furthermore, most traditional farm layouts require space for large tractor wheels to fit through rows of plants. FarmBot tracks can be placed farther apart than tractor wheel pathways and the width required per track can be narrower.

11

Figure 4.6.1. Hexagonal close packing has a density of 90.7%. Figure 4.6.2. Cubic packing has a density of 78.5%.

4.7. Eliminated Soil CompactionTraditional tractor equipment is extremely heavy and all of that weight compacts the soil each time the wheels roll over it. The use of tracks allows the weight of all FarmBot equipment to be supported by the tracks and ultimately the track supports and foundations, a space that will not be used for growing plants as the tracks are fixed in place for the long term. Eliminating soil compaction eliminates the need for regular heavy plowing and other operations that are destructive to the soil structure. This saves time and increases the health of the soil.

4.8. Continuous Land UseBecause FarmBot is able to individually tend to each plant and the section of land it is on, as soon as that plant reaches the end of its life cycle, a new plant can be put in. This allows for continuous use of all available space, independent of when crops are planted or harvested, which ones mature faster or slower, and if any plants fail to germinate or grow properly. Furthermore, plants of the same species can be planted at different times without losing machine efficiency in order to extend the season and availability of crops.

4.9. Variable TerraformingIn many agricultural practices, the soil is shaped into berms and swales of certain sizes to better suit the needs of the plants. These soil structures are formed in straight rows, usually regardless of the land’s topography. FarmBot could utilize topography maps to create soil formations and plant layouts on contour, allowing rain to be slowed and soak in rather than washing everything away, as well as reducing the need to water.

12

4.10. Open­SourceMy decision to open­source the FarmBot project was out of the notion that I do not and do not want control and ownership of the technology as well as my love for other open­source projects and the impacts they have.

I see many things wrong with proprietary technologies that stifle innovation and halt others from creating better systems. I do not want something so fundamental, like technology to grow food, to be patented and unavailable for free use. I cringe at the horror stories of companies shutting down small farmers because birds and wind have carried patented seeds onto other farms. I do not want that type of corrupting power to influence the FarmBot project. This technology is for humanity and is about feeding the world and taking care of our ecosystems, not about maximizing profit.

This does not mean that nobody will be able to create a viable company off of FarmBot technology. There will always be a need for professional, quality equipment to be designed and mass produced for greater affordability. I look at other open­source projects such as Android as a great example; anyone can use the base technology for free and then augment and modify it to be better and more unique, and then sell the value they have added. Another example is the RepRap project, where the base plans are free and open­source but many companies have refined them and now successfully sell their spin­off product.

Moreover, I see so much benefit that can come out of opening up the technology to everyone. I look to Wikipedia as the prime example of open information by the people, for the people. I can only hope that one day the community will build the FarmBot technology and database of information to as high of a caliber as the best open­source hardware, software, and data projects of today.

I will try my best to document my thoughts and FarmBot development by posting plenty of images, concisely describing my design intent, and making any CAD files and software available in multiple common formats wherever possible. All of the work from here on out will be posted on wiki.farmbot.it and I hope this will become the central place for FarmBot development.

13

5. HardwareFarmBot hardware is very similar to 3D printer and CNC milling machine hardware. Looking at Figure 5.1 for reference, you can see that there are two fixed tracks extending in the X­direction and a gantry that spans the tracks and moves along them. Mounted to the gantry is a cross­slide that moves in the Y­direction and mounted to that is the tool mount that moves in the Z­direction. Tooling includes most traditional agriculture tooling that is specially adapted for FarmBot use. The tracks, gantry, cross­slide, and tool mount design intent allow for easy scaling in the X, Y, and Z directions.

Figure 5.1. FarmBot hardware high level overview and coordinate system.

14

5.1. TracksFarmBot tracks are one of the components that differentiate FarmBot technology from traditional free­driving tractors. The tracks are fixed in the ground and allow the system to have great precision in an efficient and simple manner. There are many reasons of why tracks are superior to free­driving tractors, a few of which are listed below.

Tracks provide great precision and allow the FarmBot to return to the same position repeatedly Any type of packing structure of plants can be created and managed because wheel and hardware pathways are no longer needed Tracks take up less area than paths for tractor wheels and do not compact the soil Using tracks eliminates the need for tractor steering components and autopiloting systems

Figure 5.1.1. Tracks in relation to the other main components of FarmBot hardware.

15

5.1.1. GeometryTracks take the form of rails that are slightly elevated off the ground by supports and small concrete foundations. Each rail acts as a linear guide, providing an interface for the gantry to mechanically mate with and travel along. Each track has sufficient cross sectional area and strength to resist deflection during high force operations such as plowing. Tracks and their foundations scale in size and strength as the gantry size and number of simultaneous operations increases. Tracks may also feature a live rail to provide electrical power to the gantry and other parts.

The most basic FarmBot system needs at least two tracks in order for one gantry to span between them. A three track system can exist that allows for two gantries to operate separately on their own sections of land while sharing a middle track. Four, five, etc track systems may also exist with more gantries. Because of this scalability, there are two types of tracks: single rail, and dual rail. Single rail tracks allow one gantry to move across while dual rail tracks allow two gantries to share the same track as in the three track system.

For small FarmBot systems, the tracks could be constructed from T­slot aluminum extrusions for ease of manufacturing, flexible assembly, relative low cost, expandability, and general availability. For larger applications, custom steel tracks would likely be the material of choice for reduced cost, increased strength, and weldability. Large, pre­fabricated tracks the length of a semi truck could be shipped in and bolted or welded together on­site like railroad tracks.

5.1.2. ScalabilityAs mentioned in section 5.1.1, track systems can be scaled in the Y­direction by simply adding more tracks and more gantries to the system or by making the gantry wider. Tracks can also scale in the X­direction by making the tracks longer and adding more supports. Theoretically, the tracks can be miles and miles long in an industrial application with the only limit being the amount of area one gantry could properly tend to with the available amount of time.

Another idea for scalability is a serpentine type track system that one gantry could use, requiring curved track sections at the serpentine edges for the gantry to move to the next row of tracks. There may also exist other methods that the gantry could transfer tracks by, but these will not be covered in this paper.

5.1.3. CostThough the capital cost of any type of track system is new in agriculture, the tracks are designed to be as cost effective as possible by being simple to manufacture and lacking any moving parts. Work will need to be done to optimize track cross­sectional area and therefore material usage as well as easing the installation process. It is estimated that the up front investment of tracks can be offset by the savings from the elimination of the more complex drivetrains, steering, brakes, cockpits, and other components of tradition tractors. In addition, lifetime savings will occur from increased productivity of the FarmBot system over conventional systems simply by removing tractor driver labor and allowing for 24/7 operation.

16

5.2. GantryThe gantry, highlighted in Figure 5.2.1, is the structural component that bridges the two tracks and moves in the X­direction via an X­direction drive system. It serves as a linear guide for the cross­slide (section 5.3) and a base for the Y­direction drive system that moves the cross­slide across the gantry in the Y­direction. It can also serve as a base for mounting other equipment such as seed bays, tools, electronics, inputs, and sensors.

Figure 5.2.1. Gantry in relation to the other main components of FarmBot hardware.

5.2.1. GeometryThe gantry’s primary structure is an upside­down square U shape. At each end of the U, are linear guide systems such as wheels that allow the gantry to move across the tracks in the X­direction. The top of the U shape serves as the bridging component and the linear guide for the cross­slide.

The gantry must be very rigid and have tight tolerancing on the linear guide interfaces. Significant flex or play will lead to less accuracy of the tool or sensor location. This can be especially important during high force operations that also require high precision, such as selective tilling, where inaccuracy in excess of 1 cm could damage desired plants.

Similar to tracks, the gantry will likely be constructed from T­slot aluminum extrusions for small scale applications and welded steel for larger scales.

17

5.2.2. Drive SystemAn optimized drive system for the gantry is dependent on the size and application of the FarmBot system. For smaller systems, such as a seedling­only application, a timing belt and pulley may work the best due to low cost, ease of installation, minimal maintenance, and good precision. For larger systems, belts may introduce an unacceptable amount of slack and stretch and thereby reduce the level of precision. It also may be costly or infeasible to implement strong enough belts to handle plowing and other high force operations. In this case, a rack and pinion style drive system may work better. In this system, a stepper motor and pinion gear could be mounted to the gantry and the tracks could have geared racks mounted to them in order to mesh with the pinion.

5.2.3. ScalabilityThe gantry can scale in the Y­direction by constructing it to be wider. This modification would require the tracks to be spaced farther apart as well. The gantry can also scale in the Z­direction to accommodate taller plants such as corn, sunflowers, and even trees, by making the basic U shape taller. This modification would require a longer tool mount to be used. As with all scaling up, the structure will need to increase in strength to resist deflection and drive systems will need to be more powerful to move the increased mass.

5.3. Cross­SlideThe cross­slide, highlighted in Figure 5.3.1, moves in the Y­direction across the gantry. This motion provides the second major degree of freedom for FarmBots and allows operations such as planting to be done anywhere in the XY plane. The cross­slide is moved using a Y­direction drive system and functions as the base for the tool mount.

Figure 5.3.1. Cross­Slide in relation to the other main components of FarmBot hardware.

18

5.3.1. GeometryThe cross­slide consists of a linear slide and a mounting plate. The linear slide interfaces with the gantry while the mounting plate provides the base for the tool mount to interface with. The cross­slide must have high tolerancing with the linear slide interface and must be rigid enough to transfer high forces to and from the tool mount to the gantry without significant deflection.

5.3.2. Drive SystemSeveral drive system options exist including a timing belt and pulley, a rack and pinion system, or even a leadscrew. Each option has advantages and drawbacks and may work better than others in certain applications.

A timing belt and pulley system would work well in small applications as it is easily upgraded to longer lengths, easy to source and purchase off the shelf components, requires little maintenance, and is affordable. However, as with the gantry belt system, larger systems may introduce too much slack and stretch, reducing precision.

Rack and pinion systems would work well for small to large systems and are perhaps the best and most versatile option overall. A rack and pinion system may require specially made components that cannot be purchased off the shelf, which could be a limiting factor. However, the same components could be used for the gantry and tool mount drive system as well.

Leadscrew systems provide the greatest amount of torque and precision, but are more susceptible to damage from dusty and dirty environments. It is unlikely that FarmBots perform high force or precision operations in the Y­direction, perhaps making a leadscrew system excessive. It is worth mentioning though for consideration on smaller applications.

5.3.3. ScalabilityThe cross­slide could scale in the Y­direction, allowing for multiple tool mounts to be attached in order to complete identical operations simultaneously. This type of scaling would require a more robust gantry, track system, and drive systems to handle concurrent high force operations such as tilling. This may also put unwanted constraints on the farm design, forcing plants into a more rigid grid. However, the potential for increased operation throughput may be worth that sacrifice.

5.4. Tool Mounts

5.4.1. GeometryTool mounts attach to the cross­slide and provide the FarmBot with Z­direction movement as illustrated in Figure 5.4.1.1. Tool mounts serve as the base for attaching tools such as seed injectors, watering nozzles, sensors, and plows. They consist of a tall structural component, a drive system, and a mounting plate or area for attaching tools to.

19

Figure 5.4.1.1. Tool Mount in relation to the other main components of FarmBot hardware.

5.4.2. Drive SystemTool Mounts can be driven with various drive systems such as a rack and pinion, leadscrew, belt and pulley, electronic solenoid, or hydraulic piston. Depending on the scale of the FarmBot and the desired accuracy and speed requirements, different drive systems will be better than others. It will be important to select a system than can move heavy hardware up and down, especially during operations involving soil manipulation such as plowing or seed injecting. Furthermore, the tool mount will need to move precisely, with perhaps millimeter accuracy for seed injection. Likely the rack and pinion and belt and pulley systems will not be powerful enough, the hydraulic piston will be too complex and expensive, leaving an electronic solenoid and a leadscrew as options. However, this is only speculative.

5.4.3. ScalabilityFor a FarmBot to tend to taller plants, the gantry must be raised in order to have adequate clearance from the plants when moving in the X­direction. With a taller gantry, the tool mount must scale in the Z­direction so that tooling, such as a seed injector, can still reach the soil. The tool mount can easily scale by making the structure taller and installing an upgraded drive system.

20

5.5. ToolsTools will attach to the tool mount as highlighted in Figure 5.5.1. FarmBot will likely utilize a custom set of tooling optimized for the FarmBot, but it will generally be very similar in form and function to existing agriculture tooling. In other words, it is likely not in the scope of the FarmBot project to reinvent the plow, but it is in the scope to adapt and optimize the plow for FarmBot use. However, it is very possible that FarmBot will open the doors to new tool designs that were not feasible or appropriate to use with conventional equipment.

Figure 5.5.1. Tools in relation to the other main components of FarmBot hardware.

The following list of tooling is likely to be close to the order of development based on importance and functionality.

1. Seed injector2. Watering/fertilizer/pesticide nozzle3. Tilling implement4. Plow5. Cutter/Shredder6. Discer7. Burner8. Combine/Harvester9. Robotic harvesting arm

21

5.6. ElectronicsThe onboard electronics used will be similar to those found in CNC milling machines and 3D printers. Motors, servos, solenoids, valves, sensors, and other hardware will be controlled with a microcontroller and supplementary power supply. The microcontroller will ideally be open­source and purchasable off the shelf such as an Arduino. The microcontroller will need software that can interpret numerical control code and subsequently drive the motors and other equipment to compete operations. There will also need to be a live Internet connection for transferring numerical code as well as sensor data to and from the web backend.

5.7. Sensors“Smart Farming,” as I define it, is using data to make more informed decisions about the setup and operation of the farm. FarmBot will be able to use the following sensors and more to collect data about the soil, plants, and weather. Some of this data can be taken at many points on the farm to create data maps as explained in section 6.1.4.

1. Moisture meter2. Thermometer3. Rain gauge4. Psychrometer (humidity)5. Anemometer (wind)6. pH sensor7. Incident light meter8. Computer vision9. Hyperspectral imaging10. Mass spectrometer

22

6. SoftwareFarmBot is as much of a hardware project as it is software. Because the entire system is automated and numerically controlled, it is useless without powerful software to create the numerical code in a smart, easy, and efficient manner. This will be done with an open­source cloud based software as a service solution that allows the user to design their farm, program and control their hardware, store and view data, and use analytics tools.

6.1. Web FrontendThe web frontend will be the primary way the user interacts with and programs their FarmBot hardware. Below is a proposal of how this component of the software could function and what it might look like.

6.1.1. User InterfaceThe FarmBot web user interface should be fun, intuitive, modern, beautiful, simple, and powerful to use. The user should be able to quickly and easily access all features of the software. I envision a tabbed layout with each tab being composed of several panes that serve various functions depending on the desired user actions.

6.1.2. Dashboard TabThe dashboard tab will give the user an overview of the operation and statistics of the FarmBot system. The dashboard tab will be broken down into several panes, each offering specific types of information and control as shown in FIgure 6.1.2.1.

One pane will be dedicated to hardware information. The user will be able to use dropdown menus to pick which types of hardware and tooling they have installed and be able to add more items such as sensors when they become available. There could also be a way to quickly activate or deactivate hardware when it goes down for maintenance or replacement. This will allow the FarmBot to skip or postpone any operations that were scheduled requiring that hardware. This pane will also allow the user to set up how large their FarmBot system is by specifying the dimensions of their tracks, gantry, and tool mount. As new tracks are added to extend the system, a single number can be changed to reflect the upgraded hardware and the new farming space available.

A resources pane could take the form of an interactive graph displaying historical, current, and projected resource usage for electricity, water, seeds, fertilizers, etc. The user will be able to zoom in the graph and see more detail such as hour by hour or even operation by operation details. Furthermore, the user will be presented with filtering options to see, for example, how much water is being used on tomatoes. The graph will not only display volumetric or other base units of resources, but also their monetary cost to the farmer based on user­entered cost/unit of resource numbers.

A third pane may show system information as well as some manual control buttons. A fourth pane may show financial analysis and projection graphs of the whole operation, including expenditures on inputs, expected revenues for crops, and other factors such as maintenance and logistics cost. This graph could also have advanced filtering and analysis options.

23

Figure 6.1.2.1. Mockup of the Dashboard tab showing four panes of interactive information.

6.1.3. Farm TabThe farm tab is where the user will design the layout of the farm and control how operations are completed, ie: where, how, and when different plants are to be grown. The interface will consist of a large middle column and a left and right sidebar. We’ll call the left sidebar the Plants and Operations Toolbox which will allow the user to select plants and operations, modify the settings, and eventually place the plant or operation into the middle column. We’ll call the middle column the Farm Map which will feature an interactive and zoomable map of the farm. The right sidebar, named the Operations Agenda, will be for viewing and modifying the list of scheduled operations. Figure 6.1.3.1 illustrates a way that this page could look.

24

Figure 6.1.3.1. Mockup of the farm tab featuring the farm map in the middle and left and right toolbars.

The different functions of the Plants and Operations Toolbox are illustrated in Figure 6.1.3.2. First, the user will be presented with a way to narrow down and choose a specific plant or operation. The user may either enter a search term, browse the “My Plants and Operations” grouping or the “All Plants and Operations” grouping. Once a plant or operation has been selected, a larger image will be displayed as well as all of the editable data fields associated with that plant or operation as mentioned in section 7.1.

25

Figure 6.1.3.2. Mockups of the Plants and Operations Panel being used to search for, choose, and view a plant and its grow settings (1, 2, & 3),modify the grow settings (4), and save those settings (5).

Once the data fields have been filled, the user can then click and drag the image of the plant anywhere onto the Farm Map. When let go, a small polygon will be created with that plant’s color. Inside the polygon will be the expected date at which that plant will be harvested and the number of plants that will be planted in that region given the size of the polygon and the seed spacing selected. The user can then move and reshape the polygon as they please, as well as change the settings in the options pane for that polygon of plants.

26

The Farm Map will be zoomable and moveable and always show a units legend similar to other online maps. There will be a menu to turn on different layers such as topography, satellite imagery, and data maps like nutrient levels, moisture content, etc, as discussed in section 6.1.5.

The Operations Agenda as shown in FIgure 6.1.3.3 will show all scheduled operations that the FarmBot will complete as well as a calendar. The operations are created and modified when the user drags new plants into the farm map or edits the settings of already planted areas. The user may use the search to filter and find specific operations and can click to see details of that operation and make quick edits. Furthermore, the user may change the selected date and the Farm Map will update to reflect the predicted layout of the farm on that date. This will allow the user to plan in advance the planting of crops as soon as open space is available after a crop has been harvested as mentioned in section 4.7.

Figure 6.1.3.3. Mockups of the Operations Agenda showing different filtering and the editing process.

27

6.1.4. Data TabBy using the FarmBot hardware to take extensive data measurements, the software can create data maps that the farmer and the decision support system (section 6.2.4) can use to make smarter, data­driven decisions, resulting in optimized farm layouts and operation settings. These maps can be used to show where amendments are needed, where to water more or less, temperature, pH, etc. Extensive mapping at different times can show how the soil and other conditions change over time, allowing for decisions to be made that take into account historical data and projected future conditions. Moreover, long term trends such as yield will be easier to discover and analyze with such extensive data.

Figure 6.1.4.1 shows a mockup data map of moisture content. Assuming the data measurements were taken a few days after a medium rain, we can see where certain areas have since dried up and where other areas are still moist. Using this map, watering operations in the blue, wetter regions can be decreased or even canceled for a few more days, whereas that might not be the best decision for the yellow and red zones, depending on what is planted there. This map may allow better plant placement decisions to be made in the future, for example placing plants susceptible to root rot in areas that show quicker drying after a rain.

Figure 6.1.4.1. Mockup data map of moisture content in the soil.

28

Cost effective and extensive data mapping and analysis has the potential to be one of the most powerful and disruptive aspects of the FarmBotsystem. There has been a lot of talk and development in recent years about applying “Big Data” to agriculture. Many agree that big data is going tocreate much more efficient and sustainable production, though the current problem is that acquiring that much data requires a lot of time andexpensive equipment. Other systems being developed include flying planes and drones overhead to take imagery, driving around and manually takingsoil samples, and using tractor mounted optical sensors. FarmBot has the potential to collect more data, more accurately, more efficiently, and in amore cost effective manner. Furthermore, most of the technologies being developed are either proprietary or so prohibitively expensive to own thatfarmers must purchase the data as a service. FarmBot could make Big Data collection and analysis affordable and open to all.

6.1.5. Manual ControlUsers will be able to manually control the FarmBot through the web interface using some simple control buttons. One reason to use manual control is in the case of maintenance or troubleshooting that requires the movement of certain components. Another, would be to quickly complete simple, one­time operations.

6.1.6. User ExperienceThe user experience will be similar to some video games such as SimCIty and Farmville, where the user is playing a puppet master type role that exhibits extensive control of the landscape and its operation. All components will need to feel easy to complete, intuitive, and fun. The gamification elements like the resource tickers, graphic animations of the plants and land will add to the fun and strategy involved.

6.2. Backend

6.2.1. User ProfilesTo use the FarmBot software, one will need to create a user account so that their work and data can be saved and later accessed. The user profile will also need to be linked with any hardware that the user owns in order to provide authentication when uploading numerical control code to the hardware. See section 9.10 for thoughts on hacking.

6.2.2. Farm ProfilesEach user will need to store the data associated with their farm including size, plant layouts and settings, scheduled operations, statistics, data maps, farm history, and more.

6.2.3. Equipment ProfilesThe software will need to know exactly what hardware the user owns and its current configuration. Otherwise, the user may schedule operations that cannot be completed or the numerical control sent to the machine will not be compatible. The equipment profile information will need to be stored and easily modified as equipment is upgraded or become inactive for maintenance, etc.

6.2.4. Decision Support SystemEverything regarding the farm layout and management is a decision, such as seed spacing, plant pairing, watering amounts, where to plant and

29

when, etc. In order to optimize efficiency, the farmer or software must make smart, data­driven decisions. The decision making process can become very complex with many data streams and when data is variable such as weather patterns and soil properties.

The decision support system is an algorithmic component of the software that uses all available data to determine the best settings for every operation. This will perhaps be the most technically difficult and complex component of the software to develop and will take mountains of experimentation and expert agronomist and farmer input to optimize. The development of the decision support system can begin with only a few decisions based on only a few data sets, but the idea is to build a smart enough system such that optimum settings can be found for every plant, in every condition.

Figure 6.2.4.1 outlines a high level overview of how the decision support system could function. Essentially, the system would determine the optimum settings for every operation that FarmBot completes based on the available data streams.

Figure 6.2.4.1. The Decision Support System combines data streams and makes data­driven decisions for every operation setting.

30

6.3. Microcontroller SoftwareAll sensors, motors, and other hardware will interface with a microcontroller in the FarmBot that must interpret the numerical code coming from the backend and also send sensor and real time data back online. The microcontroller will therefore have its own embedded operating system of sorts that can interpret code, and send and receive data to the backend and to the motors and other sensors. I think it makes the most sense to do all of the more complex computing with the decision support system in the cloud backend and then send only the basic operation instructions to the FarmBot similar to the G­code used in CNC machines and 3D printers.

6.4. Data Sharing and Open Data RepositoriesTo help FarmBot owners create and manage their farms, there will be options to share and publish many different data types. I hope that an Open Data Repository will be created that can help centralize and make accessible and free all of the information needed to grow every type of plant, in every location, in every condition. Access to the open data from the FarmBot web frontend will be in the “All Plants and Operations” area of the Plants and Operations Panel. In addition, the data repository may have it’s own frontend interface for browsing, searching, and downloading the data. This will allow other applications and technologies to use the data for their own purposes as well through an open API.

6.5. Mobile ApplicationsIt may be beneficial to develop mobile applications to allow for a better user experience when programing and monitoring FarmBots. However, with responsive design of the web frontend, dedicated mobile applications may not prove to add any greater value than simply using the website on a mobile device. It also may be best to focus development resources to improving the web frontend than spreading development over multiple user interfaces. However, I understand my ignorance in this subject and will respect any developer’s reasoning to pursue mobile application development.

6.6. Open­Source and an Open PlatformAs stated before, all FarmBot software will be open­source and this is for several reasons. First, it is the essence of this project to make owning and operating a FarmBot as accessible as possible and that means making the software free and easy to use, downloadable, and modifiable. Second, the more people with the ability to contribute to the software, the better it will become, faster. I look to Linux, Android, and Wikipedia as great open­source examples. Third, open­sourcing the software allows the community at large to validate the software for security issues.

A goal from the beginning of software development will be creating an open platform that users are not locked into. The FarmBot software service should not be a “walled garden.” Instead, users should be able to easily import and export their data in accessible formats. Moreover, developers should be able to build add­ons and their own services that improve the experience and capabilities of the user. This could happen with the development of an API for data sharing across applications and services.

Moreover, users will not be required to own or use FarmBot hardware to sign up for and use the FarmBot software. The software can still help any farmer or home gardener plan, keep track of, and better manage their farm or garden.

31

7. DataAgriculture as a method is successful when knowledge or data is used to make decisions that have been proven successful before. FarmBot’s ability to obtain and combine thousands of data points, plant knowledge, and other factors in it’s decision support system allow the farmer to know that their farming operation is data­driven and proven. Sections 7.1 through 7.8 describe potential sources of data that FarmBot can utilize, with the ultimate goal being complete automation of the farm design and operation processes in an optimized way.

7.1. Plant DataPlant data includes everything that the FarmBot needs to know in order to successfully plant, grow, and harvest a specific variety of plant in the best possible way. Below are some of the parameters that might be included in plant data.

Full sun, partial sun requirements Seed spacing Seed depth Number of seeds per hole Type of planting foundation (mound, hole, none, etc) Optimum watering regimen throughout plant life (volume, timing, and location) Optimum fertilizing and spraying regimens throughout plant life (volume, timing, and location) Special conditions such as frost susceptibility, root rot resistance, etc Optimum soil conditions such as nutrient levels, soil composition, drainage ability, etc Best companion plants Beneficial and problem insects attracted

7.2. Soil DataAccurate, plentiful, and updated soil data is perhaps the hardest, most time consuming, and expensive data to attain with today’s methods. Soil type and conditions can change quickly both spatially and temporally. For example, after a deep till, a heavy watering, or a nutrient intensive crop is grown, soil conditions will have changed. Geological features such as streams, valleys, mounds, and higher concentrations of rocks or clay can change soil conditions every few feet. FarmBot can very quickly take many data points in many locations in a systematic way and create high resolution maps of this data. These maps enable the farmer and the decision support system of section 6.2.4 to make smarter, data­driven decisions. Types of data a FarmBot could gather about the soil include the following and likely many more.

Percent organic matter, clay, rock, sand, etc Moisture content

32

Percolation rate Nutrient content Microbiological activity Temperature Incident light levels pH Alkalinity Electrical conductivity

7.3. Companion Plant DataA key component in farming a polycrop is choosing good companion plants. A great example is the Three Sisters system of the Native North Americans, combining maize, squash, and climbing beans. From Wikipedia: The maize provides a structure for the beans to climb, eliminating the need for poles. The beans provide the nitrogen to the soil that the other plants utilize, and the squash spreads along the ground, blocking the sunlight, helping prevent establishment of weeds. The squash leaves also act as a "living mulch", creating a microclimate to retain moisture in the soil, and the prickly hairs of the vine deter pests. FarmBot enables the farmer to create a farm layout of any complexity without sacrificing machine efficiency. Using best companion plant data associated with every plant, the FarmBot software could suggest the best combinations to plant together. This data, combined with known insect troubles, nutrient deficiencies in the soil, the time of year, and other factors, will allow the farmer or the software itself to generate optimum farm layouts.

7.4. Time and Location DataCurrent time and location data can allow the FarmBot software to determine daylight and nighttime hours in order to perform certain operations at optimum times such as watering or putting seeds in the ground. Location, orientation, and date data will enable smarter plant layouts based on how high or low the sun will be and at what angle it is in relation to the land and plants.

For example, shorter plants can be planted closer to the sun in order to not block sunlight from farther plants. This concept would be more prevalent during months when the sun is lower in the sky but not as important when the sun is higher as well as on sloped land.

7.5. Weather DataThe FarmBot software will eventually plug in to local weather data and forecasting for the following potential benefits.

Water plants in a more efficient manner based on rain forecasting and daily temperatures Protect against frost based on temperature monitoring Ability to start plants at the best times based on the upcoming forecast Create farm layouts that are more resilient against high winds

33

7.6. TopographyTopography plays a major role in how a farm is planted, how sunny areas are, and where the water wants to flow. Using topography data, the farmer and software can place plants in better locations and also create more effective landforms such as on contour lines rather than straight lines. Furthermore, the software will be able to better predict and verify with data the flow of water and nutrients after heavy watering or rains.

7.7. Past and Future DataFarm and weather history and future planning will play an important role in optimizing farm layouts in the following ways.

Prior success or failure of certain plants or sections of land including quantitative measurements such as yield Prior planting history and soil quality growth or degeneration Planning for more optimized growing of future crops (building the soil now as an investment) Planning for market price fluctuations for inputs and the end products

7.8. Manual Input DataThe farmer will always be able to manually change variables and input special pieces of data in order to have complete control over the farm design. These manual inputs may be overriding, modifying, or adding to the existing data. For example, if the farmer knows a herd of cattle is going to graze a certain section of land, there will need to be a way to account for that within the software through special pieces of manually input data.

7.9. Downloadable and Auto­generated FarmsEventually, I hope that the FarmBot community will create many different farm and garden layouts that can then be easily downloaded to any FarmBot. This will allow the layperson to grow an optimized farm without as much experimentation, research, or prior knowledge. Even farther down the line, FarmBot should be able to be installed, turned on, take data, check all other data sources, and then auto­generate an optimum farm.

34

8. FarmBot GenesisFarmBot Genesis, currently in development and rendered in Figure 8.1, is the first FarmBot that will be built to a functioning state. It is designed to be a foundation for experimentation, prototyping, hacking, and learning, and will initially perform only two operations: injecting seeds and watering them, with the intention that more features, tooling, and functionality can be added later. The driving factors behind the design are simplicity, manufacturability, scalability, and hackability.

Figure 8.1. Rendering of a 1.5m wide, 3m long, 0.5m tall FarmBot Genesis.

35

Genesis is a small scale FarmBot primarily constructed from V­Slot aluminum extrusions and hardware from Open Builds (openbuildspartstore.com.) The V­Slot extrusions were chosen because they are easily cut, the structure can be quickly assembled and modified, and each extrusion can also function as a linear guide. Stepper motors and an Arduino are used to control the system and were chosen for their availability and hackability. Genesis can vary in size from a planting area as little as 1m2 for seedling­only applications to greater than 50m2, while accommodating a maximum plant height of 1m. With modifications to some of the structural component sizes and an alternative X­direction drive system, Genesis could scale up to a 1000m2 planting area and a maximum plant height greater than 2m.

8.1. Genesis TracksGenesis tracks, rendered in Figure 8.1.1, are designed using V­Slot aluminum extrusions and hardware from Open Builds. The V­Slot extrusions function as the linear guides for the gantry to move across and the design intention allows for easy scaling in the X direction. The tracks may be fixed to the ground by attaching them to concrete pilings, burying them in the ground, fastening them with large stakes, or other methods. Once installed, it will be very cost effective to scale in the X­direction by simply adding more track sections. If scaling in the Y or Z direction is desired, it may require upgrading the main track beam to a larger size in order to reduce flex under the increased loads.

Figure 8.1.1. Rendering of a single Genesis track and X­direction drive system.

36

8.2. Genesis X­Direction Drive SystemThe Genesis X­direction drive system, partially seen in Figure 8.1.1, is a dual belt and pulley system consisting of Nema 17 stepper motors, GT2 timing belts and pulleys, and mounting hardware from Open Builds. One stepper motor and belt system is attached to each track, with the belt ends fixed to the gantry for actuation. To scale in the X­direction, a longer belt will be required. Nema 17 stepper motors and GT2 components were selected for their common use in DIY 3D printers, affordability, and general availability.

8.3. Genesis GantryThe Genesis gantry, rendered in Figure 8.3.1, is constructed of V­Slot aluminum extrusions, mini­V linear guides, other components from Open Builds, and some custom support braces. The gantry is primarily designed to be rigid while still allowing for easy scaling and places to mount other hardware. The Genesis gantry interfaces with the tracks (Figure 8.3.2) and also provides the linear guide for the cross­slide as seen in section 8.5.

Figure 8.3.1. Rendering of the Genesis gantry.

37

Figure 8.3.2. Detail rendering of the gantry interfacing with a track.

38

8.4. Genesis Y­Direction Drive SystemThe Genesis Y­direction drive system, rendered in Figure 8.4.1, is very similar to the X­direction drive system; it uses the same Nema 17 stepper motors and GT2 timing belt and pulleys. It is also scalable in the Y­direction with a longer belt.

Figure 8.4.1. Detail rendering of the Y­direction drive system mounted to the gantry and interfacing with the cross­slide.

39

8.5. Genesis Cross­SlideThe Genesis cross­slide, rendered in Figure 8.5.1, consists of the universal mounting plate, two spacer blocks, and eight V­wheel kits from Open Builds for interfacing with the gantry (Figure 8.4.1) and the tool mount (Figure 8.6.2.)

Figure 8.5.1. Rendering of the Genesis cross­slide.

40

8.6. Genesis Tool Mount and Z­Direction Drive SystemThe Genesis tool mount and Z­direction drive system, rendered in Figure 8.6.1, consist of a single V­Slot extrusion with a Nema 17 stepper motor, leadscrew, and mounting plates. Both the extrusion and the leadscrew interface with the cross­slide as shown in Figure 8.6.2. A leadscrew was chosen over a belt and pulley for its ability to sustain a load without using motor power, as well as its increased torque and precision which will be necessary for precision operations like seed injection.

Figure 8.6.1. Rendering of the Genesis tool mount and Z­direction drivesystem.

Figure 8.6.2. Detail rendering of the tool mount and Z­direction drivesystem interfacing with the cross slide.

41

8.7. Genesis ToolsFarmBot Genesis will initially be built with only the two most essential tools: a seed injector and a watering nozzle. These two tools can be modified and expanded upon for more functionality, such as a fertilizing ability built into the watering nozzle. The Genesis tool mount and microcontroller will be able to accept more tools and sensors for future expanded functionality.

8.7.1. Genesis Seed Injector and Seed BayThe Genesis seed injector, rendered in Figure 8.7.1.1, consists of two main assemblies: a 12V vacuum pump and selector tip, and the seed bay. The vacuum pump and selector tip are attached to the tool mount and allow for the precise selection of seeds from the seed bay and placing them at a precise location and depth in the ground. This is completed by first dipping the selector tip into a seed bin while the vacuum is on, moving the suctioned seed or seeds to the desired location, driving them into the ground to the desired depth, releasing the vacuum, and repeating. The seed bay, attached to a vertical column of the gantry, moves in the X­direction such that different seed bins can be positioned for the selector tip to choose from. The Genesis seed injector uses two controlled components: the vacuum pump, and the stepper motor for the seed bay. This setup will allow any number of different seeds to be accessed and planted without the need for additional controllable components.

Figure 8.7.1.1. Rendering of the Genesis seed injector and seed bay.

42

8.7.2. Genesis Watering NozzleThe Genesis watering nozzle, rendered in Figure 8.7.2.1 alongside the seed injector, consists of a spray nozzle, a solenoid valve, and tubing. Not pictured is additional tubing that will need to run along the tool mount, gantry, and tracks in order to connect to a water source such as a garden hose. Future watering nozzles may feature flow meters, different spray nozzles, or even the addition of a peristaltic pump and tank that would allow for the precise addition of liquid fertilizer or pesticide to the water spray.

Figure 8.7.2.1. Rendering of the Genesis watering nozzle (to the left and below the seed injector.)

43

8.8. Genesis Electronics

8.8.1. MicrocontrollerAn Arduino Mega microcontroller, pictured in Figure 8.8.1.1, will be used to control the stepper motors, vacuum pump, servo, and future electronics and sensors. This platform was chosen for it’s low cost, general availability, hackability, expandability through shields, the expansive learning resources available, the strong DIY community already using the platform, and the fact that it is open­source. In addition, Arduino programs are written in the C language and therefore very familiar to many. Expansion shields likely to be used will include wifi, a RAMPS stepper driver, and an SD card shield. The firmware to be installed will likely be forked from an existing 3D printer G­code interpreter and then modified for the FarmBot application.

Figure 8.8.1.1. An Arduino Mega microcontroller.

8.8.2. Stepper MotorsThe Nema 17 stepper motor shown in Figure 8.8.2.1 has been chosen for it’s general availability, common use in similar projects such as the RepRap 3D printer, easy setup and control, as well as it’s accuracy, speed, and torque outputs. In addition, this motor interfaces with components such as pulleys and mounting plates available from many providers including Open Builds. It has yet to be determined if the Nema 17 motor size will be powerful enough for FarmBot Genesis due to the larger scale of the hardware than 3D printers.

44

Figure 8.8.2.1. Nema 17 stepper motor available from Open Builds.

8.9. CostTable 8.9.1 shows a rough estimation of the cost to produce the 1.5m wide, 3m long, 0.5m tall FarmBot Genesis rendered in Figure 8.1 by purchasing as many components as possible. The prices for the components are at a non­bulk rate and there are very likely more affordable suppliers available. All of the mounting plates and brackets are also easily manufacturable, which would further drive the price down if not purchased.

Table 8.9.1. Rough cost breakdown to produce a FarmBot Genesis kit.

Component/Description Quantity Total Cost

20x20mm V­Slot Aluminum Extrusions 1.7m $17

20x40mm V­SLot Aluminum Extrusions 9.5m $124

T Brackets 12 $56

45

Corner Brackets 4 $5

Custom Large T Brackets 2 $20

Motor Mount Plates 5 $35

Idler Pulley and Threaded Rod Plates 5 $35

V­Slot Universal Plate 1 $12

V­Slot Spacer Blocks 2 $8

Mini­V Plates 5 $45

Mini­V Wheel Kits 20 $65

Dual­V Wheel Kits 8 $31

Nema 17 Stepper Motors 5 $100

Solenoid Valve 1 $8

Vacuum Pump 1 $15

Flex Couplings 2 $10

Leadscrew 1 $30

GT2 Timing Belt 15m $100

GT2 20 Tooth Timing Pulleys 4 $20

T Nuts, Bolts, and other hardware $80

Miscellaneous Items $150

Arduino Mega 1 $60

Ramps Stepper Shield 1 $25

Labor Hours (for a kit, not assembled) 5 $250

Total $1301

46

8.10. Crowdfunded LaunchI have identified two main benefits to launching FarmBot Genesis on a crowdfunding platform. The first, is the great exposure and press potential. Crowdfunding has become very popular in recent years with both open­source and 3D printer projects leading very successful campaigns. Many of these projects are covered by popular tech blogs and shared virally on social media, leading to large interest.

The second reason is that crowdfunding is a great way to quickly get a technology into a bunch of people’s hands. For FarmBot to progress quickly, a large community needs to be created to begin tinkering with the technology and improving it. For those that just want to have a FarmBot shipped to them rather than sourcing all the parts and doing a more complex assembly, the crowdfunding rewards will be a good option. Anyone else who is tech savvy and a tinkerer can just use the wiki instructions to build their own FarmBot which is also making progress.

8.11. TimelineTable 8.11.1 is a rough timeline for the development of FarmBot Genesis.

Table 8.11.1. Timeline of FarmBot Genesis development to a crowdfunded launch.

Dates Milestone

September 2013 Prepare the wiki, blog, and social media profiles for the public.

September 2013 Publish the FarmBot white paper and gather online and local interest through targeted distribution of the paper.

September 2013 Secure funding for the project in order to pay for prototyping materials and time.

September 2013 Pitch the project to Cal Poly students and professors beginning their senior project and capstone courses to get students working on the technology.

October 2013 ­ April 2014 Construct, test, and iterate FarmBot Genesis while documenting the process.

April ­ June 2014 Crowdfunded launch of FarmBot Genesis.

47

9. Potential Impacts, Ramifications, and Concerns of FarmBot

9.1. Increased Production EfficiencyIf FarmBot proves itself to function as described, it will undoubtedly help increase food production efficiency by making the current systems and methods obsolete. FarmBot has the potential to be more effective at growing crops, use less inputs and labor, and be more economic. Part of the vision of FarmBot is to grow food for everyone, and I think that is possible.

9.2. SustainabilityThe majority of today’s industrialized agriculture practices are unsustainable both in their energy and resource usage but also in their destruction of soil health and food quality. FarmBot aims to change this paradigm by offering a way to produce sustainably without sacrificing efficiency or increasing cost. In fact, the FarmBot system will increase efficiency and decrease cost. Additionally, the decentralization of food production (Section 9.4) will allow great savings in the energy and infrastructure required to ship and distribute food.

9.3. Democratization of Food ProductionFarmBot’s scalability allows anyone with access to a computer and a yard, some land, or even a roof to be able to install and operate a FarmBot. By open­sourcing the technology as well as designing for simplicity, hackability, affordability, and accessibility, FarmBot will lower the barrier to the average person to grow food. Ideally, farming knowledge and special skills will never be needed to install and operate a FarmBot.

By making the choice to grow food accessible, FarmBot will help to democratize food production, giving the power of food back to the people. No longer will food production practices, availability, and prices be determined by the food producers, the government, or the economy.

9.4. Decentralization and Localization of Food ProductionWIth more people growing food in smaller operations, food production will be partially decentralized and localized. This will make each community and even household more independent and resilient to crop failures and other large scale disasters. It will give more people access to fresher food and provide greater abundance for all. In addition, it will save tremendously on shipping cost and infrastructure requirements.

9.5. Elimination of the FarmerPerhaps the most controversial aspect of the FarmBot vision is the eventual elimination of the farmer. Historically, putting a good amount of time and effort into feeding yourself and your family is something that every living thing had to do. But since the dawn of agriculture, humans have been improving methods and technologies in order to spend less time on this task. FarmBot is perhaps the last technological advance needed to make our time spent farming go to zero.

48

By creating a mechanized system where very specific and routine actions are needed to effectively produce food, it only makes sense to hand off the remaining repetitive task to a computer. Like so many technologies in the past, this advancement could make many people’s jobs obsolete.

Depending on who you talk to, this is either good or bad. I see it as a good thing; I would rather have farm laborers becoming educated and working in other industries rather than picking berries all day, hunched over in a field. That’s not to say that the people who enjoy working on a farm and in the garden, myself included, cannot continue to do it, but now we can do it as a hobby and not necessarily because it pays the bills.

Depending on the adoption rate of a disruptive technology like FarmBot, there could be a large spike in unemployment, or a gradual fade out of farm jobs. Either way, there may be a whole new class of jobs created such as professional FarmBot installers, maintenance specialists, and anyone working to manufacture the components.

9.6. Increased or Decreased Separation of People and FarmingIt is hard to say if FarmBot will further separate or bring together people and their food. In one case, many small scale FarmBots are implemented, making growing your own food more accessible. This would allow many people to get hands­on experience, be physically close to, and have some level of responsibility and commitment with their food’s production. In the other scenario, FarmBot is adopted primarily on a larger scale which leads to even less people being in charge of producing food and those people simply sitting behind a computer, with no physical connection to the process.

But with the current trends, scenario two is happening anyway and it leaves no opportunity for scenario one. So, at the least, FarmBot has the potential to connect people with their food more than today’s technology simply because of the scalability and potential for home installation and operation. As a further matter, there are benefits to be realized with both systems, that will likely ultimately lead to adoption on both scales.

9.7. Greater Dependence on Machines and ComputersIt can be argued that humanity’s increasing dependence on technology such as machines and computers could pose serious threats to humans in a large scale crisis or infrastructure failure. This is perhaps a fact we must accept as we adopt any new technology, but it is important to step back and be mindful of our dependence on these systems, especially when the system is a means to feed ourselves.

9.8. Loss of Individual and Generational Knowledge and the Gain of Universal and Accessible KnowledgeWith the adoption of a technology like this, there may be a great loss of individual and generational knowledge in how to grow food and work with the land and natural systems. As with any knowledge or cultural loss, it is likely a choice we must make if we are to adopt new methods of doing things.

With the loss of individual and generational knowledge, we will gain universal and accessible knowledge by cataloging all of our food growing techniques into one place and making that knowledge accessible. The losses and gains of our knowledge in farming will be similar to the losses and gains that have occurred with the rise of the Internet, mass production, globalization, and other areas.

49

9.9. Lock­in to an Inferior SystemIt can be argued that other systems of growing food such as permaculture or hydroponics are the best solution to sustainably growing food for the world in an economic way. However, because of the rigged market that favors conventional growing, conventional growing is the dominating practice and we have been somewhat “locked­in” to this inferior system. If FarmBot rises above conventional as a better system, it will be important to keep our minds open to other, even better systems. We must avoid forces that could unfairly favor FarmBot, locking us into another inferior system.

9.10. HackingThe fact that FarmBots will be connected to the Internet leaves a huge possibility for hacking. Because many commercial and industrial FarmBots will one day be equipped with tools such as burners and potentially harmful chemicals, there is a large risk for a malicious person to set fire to fields and also poison the land, crops, and water supply through inappropriate practices. Furthermore, almost all FarmBots, regardless of size, will be equipped with plows and other tilling implements that could destroy crops. Lastly, every FarmBot will at least be connected to a water supply, allowing for detrimental overwatering and the waste of a valuable resource on a massive scale.

All of these security threats could cause destruction of crops, damage to the land and environment, the waste of resources, and damage to the FarmBot equipment itself on a potentially worldwide FarmBot security breach. Though I am no expert in computer and hardware security, I do know that each risk will need to be handled appropriately through various methods. Here are a couple of ideas.

Requiring secure encryption between the communication of every FarmBot and any web service Authentication requirements for any operation or firmware uploads. This could require knowing a unique serial number on the FarmBot

microcontroller or other hardware Manual valves and other hardware that require a present human to activate, preventing dangerous and unwarranted computer controlled

operations such as burning Hosting the FarmBot software backend locally and connecting to the hardware on a secure, local intranet

Because the software will be open­source, it may be checked for integrity by the community at large. On top of this, any user will have the freedom to download and host the backend software on a cloud service of their choice or their own hardware. The software may then also be modified to require higher levels of authentication or other, additional forms of security if desired.

9.11. Failure of Supporting InfrastructureBecause FarmBots may rely heavily on an Internet connection and the electric grid, failure of these infrastructures could cause detrimental effects to fields. These risks could be mitigated with the use of local hardware running FarmBot software connected via an intranet, and on­site power generation with photovoltaics, wind, or backup generators.

50

10. Defining Short Term Success and Next StepsI believe the FarmBot Project has the potential to revolutionize the way humanity produces food both on the small and large scale. As the vision states, the project aims to create an open and accessible technology enabling everyone to grow food and to grow food for everyone. However, revolution will not be the defining metric of success in the short term. I think that short term success boils down to achieving two important milestones.

The first milestone is to create and demonstrate a functioning minimum viable product such as FarmBot Genesis. This will right away prove or disprove the viability of the technology. This milestone also includes making available the plans, source code, and even purchasable kits of the minimum viable product in order to lower the barrier to entry for others to learn about and contribute to the project.

This leads me into the second milestone: creating a vibrant and excited community of DIYers, makers, hackers, and enthusiasts who will contribute to the development of the project. I don’t know exactly how this would be measured but perhaps it is reaching a critical mass of people such that the project and community is self sustaining and grows on its own.

The next steps in the development of the FarmBot project are completing the initial working hardware and software package that is FarmBot Genesis and preparing the technology and wiki for the crowdfunded launch. For this to happen in the most efficient manner, I am looking for more people to join the development team and for investment to purchase prototyping hardware and pay for the development cost.

I am looking for great people with knowledge in mechanics, manufacturing, robotics, mechatronics, microcontrollers, agriculture, botany, hydrology, soil science, marketing, graphic design, and software development including web backend and frontend, internet of things, wikis, mobile app development, computer­machine interfaces, UI, UX, algorithm development, big data analysis, etc, you name it. Everyone who is interested in the project is welcome to join the team as all skills and perspectives are appreciated. Please email rory@farmbot.it or call (678) 321­7679 if you are interested! You can also follow the blog at blog.farmbot.it, contribute to the wiki independently at wiki.farmbot.it, and follow the social media profiles.

10.1. List of Action ItemsThe following list is an incomplete, high level overview of things that need to happen for the minimum viable product and for the project as a whole.

Secure $3,000+ investment/donations Finish CAD work and sourcing parts for FarmBot Genesis Document FarmBot Genesis development on the wiki Purchase materials and construct the prototype Conduct testing and iteration of the prototype and document all experience Develop or modify an Arduino sketch to interpret G­code, drive the hardware, record data, and communicate with the backend

51

Develop a backend that supports user profiles and data, remote connection with an Arduino, and has a decision support system Complete the design of the web frontend Design graphic assets including buttons, plant icons, logos, etc Develop the interactive and responsive web frontend Develop or set up an open data repository and API that is searchable and accessible by people and the backend Aggregate plant data in the open data repository Plan for the crowdfunding launch Tons of other things

52

11. ConclusionsThank you for reading this paper and learning about a project that I am passionate about and have dedicated a lot of time to. I believe this idea can bring us closer to sustainable food production for all and I hope you think so too. If you found this paper interesting or know someone who might, please pass it along through that series of tubes where you came across it.

Please send any feedback or inquiries about the technology or joining the development team to rory@farmbot.it or call (678) 321­7679. You can also follow the blog at blog.farmbot.it, contribute to the wiki independently at wiki.farmbot.it, and follow the social media profiles.

­ Rory

53