by Juan Carlos del Rio, Giulia Klein, Alix McCabe, Scott Mackinley, and Brennan Vandenhoek
We were tasked with creating a “FarmBot”: a robotic system that can plant, cultivate, and harvest crops. Our design was created in response to a specific set of design constraints. From there we were inspired to pursue a non-traditional CNC planting configuration, namely a spiral. Our final assembly can be broken down into five major subassemblies: post, chassis, seeder/waterer, weeder, and harvester. What follows is a brief overview of each of these sections with more detailed design descriptions later in the report.
The guiding force for this project was the extremely low budget constraint that we were given for the size of plot that we were tasked with cultivating. In response, our team brain stormed ways to keep costs low while keeping a large cultivation area. Here is our logic:
- Cost is proportional to weight
We can decrease weight by reducing the aspect ratio of our plot. The gantry that needs to access any point in a rectangle can weigh less if the gantry is skinny and simply rolls a longer distance along the ground.
The logical extreme of this aspect ratio reduction is a line where a small “gantry” (essentially just a cart) rolls down a long row.
Having a 350’x1’ garden is inconvenient, but it can be made more compact by spiraling it inward along an Archimedean spiral. Such a figure is easily constructed by tethering our cart chassis to a post and having it wind itself in or out to access any point in the plot. This proposition also provides a whole host of benefits such as: our plot size is easily augmented by increasing the tether length, position within our plot is reached with a single motor, and resolution is adjustable from the diameter of the internal post. Furthermore, our design is mobile. If needed, one can simply plant an additional post and use the same machine to cultivate a separate plot without the need for heavy moving machinery and equipment.
Two key issues that as a design team we had to always keep in mind were cost and robustness. These were especially hard to satisfy as it is hard to maintain strength when minimizing costs. However we believe that the platform that the chassis provides was a great compromise between both variables. For us as a team it was also really important to keep the chassis as simple as possible so that farmers would be able to fabricate the entire thing on site from tube stock.
The design seen above is a welded construction made out of 1 in SQ x 0.065 A5130 Mild Steel tube. We leveraged the use of tubes as much as possible for mounting in order to eliminate the need to have milled or waterjet parts that would make it harder for the farmer to repair on site without specialized equipment. The chassis is powder coated in our design (though it can also be spray painted) in order to protect the tubes from the environment, this in addition to proper storage over the winter would ensure longevity for the robot.
Seeding and watering Mechanisim
The seeder was inspired by hydro seeding and power washing - two high-pressure water-driven systems. It is driven by a compressor that forces water through a high-pressure tube, and delivers seeds to the jet stream via six independently actuated seed canisters. These canisters are opened and closed with solenoids. It’s fundamentally based on three base plates made of either aluminum or steel that fully retain the system and enable ease of mounting to the chassis. The seeder is also capable of spaying just water. We are using the same mechanism, run at a significantly lower pressure, to water our crops; hence providing precise and efficient water distribution.
The harvester uses two knurled aluminum cams to passively grasp plant stalks. This mechanism was inspired by jumar mechanisms which allow climbers to clamp onto rope when it is moving in one direction and allow the rope to pass through easily in the other. Similarly, the pair of cams are mounted on off-axis axles which culminates in their spacing decreasing as they rotate inwards. This motion lets our harvester hold on tightly to plants when uprooting them, but when force is applied in the opposite direction, the plants are able to slip free. Such a loading scenario occurs when the harvester is rotated upwards to dump the plant material into a tub located on the chassis. Because safety was a concern, it is significant to note that the harvester grasps plants passively which means any sentient being would easily be able to free themselves from the cams without harm. Also, the harvester subassembly meets the requirement of being cheap because it is made from few parts and all of which are inexpensive and easily replaced.
Seeking to reduce the amount of cost and complexity of our system for weeing we chose to use a pair of propane torches. We chose this for a number of reasons. First, propane torches are a very environmentally friendly way of managing weeds. It does not disrupt the soil, it just wilts the weeds and keeps all the nutrients in the soil. It also allows for no need of pesticides and reduces the need for fertilizer.
The weeder has no moving parts. This means that there fewer points of failure. Making the system incredibly durable and reliable. Further all of the parts can be bought off of the shelf. This makes it very cheap and easy to build pretty much anywhere. Also, it makes it a lot easier to repair as if anything breaks it can easily be replaced for very little work.
The post is used as a center point to ground the farmbot and provide its path of motion. It is a simple cylindrical design that is made out of concrete so that it can be poured directly into the ground using a cement tube. At the top of the post there is a fixed spool that the farmbot is attached to by a cable, this leads the farmbot to move in a spiral path towards the center maximizing use of land.