Define
The purpose of this project is to create a functional, solar powered, automated, biofuel harvester for the crop Camelina, working in conjunction with Kuilima farms on the North Shore. Originally, this project was started as a passion project for the Punahou Sustainability Fellowship, but in order to properly access the manufacturing tools, and to best use my time, I decided to work on this project as my main Engineering 3 project.
This project has the following requirements:
Continuous harvesting of crops for periods of an hour or longer
Collection of harvested crops
Autonomous Operation
Collection of Crops
Operation in uneven and rough terrain (however, lacking hills)
This project has a predecessor that was built during my time in Engineering 1 and 2. However, there is a major shift in the objective, shifting from the planting of seeds, to the harvesting of seeds. Additionally, with my increased experience with CAD, especially in the setting of technical robotics, myself and my partner are much better equipped to design and build such a machine, creating a much better design.
Purpose Document is Here: https://docs.google.com/document/d/10fkYtHBYmrvEaAI1Uflzeu8xNM5dXDt1qdnU7Igv_Wo/edit?usp=sharing
Inspiration
As previously mentioned, this project has a predecessor that comes in the form of a Solar Powered Seed Planter that myself and my partner worked to build in Engineering 1 and 2. This design uses many of the successful practices accomplished in the previous design, mostly in the form of drivetrain, and learns from the mistakes and failures of the previous design, to create something robust and sturdy.
I've always been extremely passionate about sustainability; I believe that creating a more sustainable society, a more sustainable world, is one of the most important things for humanity to achieve in the upcoming years. With this passion and inspiration of mine, this past summer, I have been a member of the inaugural cohort of Punahou Sustainability Fellows. With this opportunity comes a responsibility to create a more sustainable world, and to use my resources, connections, knowledge, and skills to make the biggest contribution that I can towards this effort.
Over the course of this past summer, myself and my fellow fellows traveled around the island, visiting different companies and organizations both dedicated to, and not, to sustainability; we visited companies in every sector of sustainability, from the energy sector (in the form of HECO, H-Power, Mililani Solar Farms, and many other locations), to the local agricultural sector (in the form of local farms and food importation facilities), to the transportation sector (in the form of interviewing leads of Hawaiian Airlines, and visiting transportation centers). From this experience, I learned the importance of biofuel development, and even the efforts that are going on here in Hawaii. From this, I learned that it is a grueling manual task that people must perform to harvest these biofuels on island, and the alternatives are large, gasoline powered pieces of equipment; quite obviously, neither of these methods are sustainable (both for the people, the economy, and the planet).
Thus, I thought it was necessary to create an automated piece of equipment. This will both create an easier harvesting process, and allow for decreased carbon emissions in this sector of energy production and biofuel research and development.
For this project, I have had the great opportunity to work in conjunction with Kuilima Farms, which is a local farm based up on the North Shore. This farm is of particular interest because it introduced myself and my fellows to the biofuel crop which this harvester is created for, Camelina. They are working with us to provide us the necessary information to build this project, and will eventually also allow us to test our prototypes on the crops that they have available, an exciting opportunity for inexperienced high school students.
While this robot is specifically being built to harvest Camelina, we are hoping to build our header in a way which makes it modular, and able to switch between different types of crops in an extremely convenient manner.
I understand that it may not be extremely clear how this project is related to FTC Robotics, but this is a legitimate goal for this project, and the process that I am going through to create this project.
I am currently on the team 6962 Pokebolts for FTC robotics, which has recently won the Hawaii regional to advance to worlds. However, I feel that my contribution to the team could be greatly improved for the next season, and I use project such as this to improve my CAD and design skills. Additionally, through this project, I am hoping to improve my programming ability, potentially even becoming a contributor on that end of my FTC team.
Ideation
This project requires a heavy and sturdy drive train base, which requires a lot of torque. For this, we believed it best to create a differential tank drive system, as I have a significant amount of experience when it comes to creating differential drive systems.
This system would use a primary gearbox that acts as an engine equivalent, with 6-10 CIM motors in this gearbox, which drives the robot. There would be a secondary motor gearbox, consisting of 2 high performance Kraken robotics motors, which would act as a differential mechanism.
A key part of this design is building a thresher which is capable of separating the usable organic mass from the unusable stalk and chaff. For this, my partner and I decided to create a system which uses a rotating drum with a mesh screen to sift through the extremely small seeds, which are the only usable part of Camelina.
This is powered by a single, geared up CIM motor, which should produce the necessary power that is required to drive the drum, and shake out all of the seeds.
Another one of the extremely important parts of this project is creating a header (something which "heads" the robot) capable of cutting the stalk and properly intaking the raw plant mass into the robot. For this, we decided to use a series of funnels and rotating blades to cut at the stalk of the Camelina, and a series of back-funnel augers that funnel the harvested crop into the intaking point.
While not similar to existing designs that are used on similar crops (most of these use oscillating blades), this design was heavily inspired by the designs used on major food crops, primarily corn and grain.
Another major goal of this header is to create something which is interchangeable, depending on the crop that is being harvested. A modular design enables quicker repairs, and greater flexibility of the robot in its deployments.
This design takes major inspiration from the common "parallel-plate" design of many FTC robots. Instead of using traditional mechanum wheels, this robot is using treads sandwiched between two parallel plates, the inner being a major structural component (cut from 1/4" aluminum), and the outer being a deflection panel (cut from 1/4" acrylic).
This design utilizes 2"x1" aluminum channels as a primary structural component, having the plates stabilized and properly aligned using the aforementioned channels.
CAD Iteration
Iteration 1 (of many)
The primary goal of our first iteration was to create a viable prototype for a differential drive system, which would be both sturdy and reliable. This first iteration aimed to create a visual ideation of what a differential drive system could potentially look like, and what mechanical systems would need to be developed to create a system with the aforementioned properties (sturdy and reliable). However, this first version was never meant to be a final prototype, or even turned into something with properties similar to a final prototype - it was merely a learning experience. For this first iteration, it was primarily focused on the development of the internal gear mechanisms.
We realized that this design had many major flaws, some of which we ignored in our next prototype (we heavily regretted this), some of which we acknowledged and designed to improve upon.
The first of the flaws that we acknowledged was our extremely small gear modules, something that was sure to break under the immense loads that this module would be placed upon. Secondly, our couplings and shafts, specifically on our large drive shafts, didn't meet any industrial standards, and would have to be custom machined - thus, this should be changed to be something that can be COTS purchased. Finally, the assembly of the internal gear, as we found, was extremely tedious at times - specifically with the assembly of the outer gears, the threading, and the insertion of nuts into extremely small spaces.
The first of the flaws that we did not acknowledge was that many of the parts were extremely hard to manufacture in shop, and would either have to be 3D-printed out of exotic and expensive materials, or would have to be outsourced to be manufactured in a professional machine shop (as they required 5-axis milling, but could actually be prototyped). Secondly, we refused to acknowledge that the differential system added a lot of mechanical complexity, and had many intricate, hard to access points of failure, which could lead to catastrophic breakdowns, and extremely difficult repair processes (neither of which are desirable). Finally, the differential system had zero potential benefits that we could utilize - we had some concerns regarding the torque required of the secondary gearbox, and the speed required of both gearboxes, and realized that it was going to be a better solution to use a differential steering system, rather than a differential gearbox (a gearbox dedicated to each side of equal power). However, as we chose to ignore these problems, we began our next iteration of our drivetrain still pursuing a complex differential gearbox system.
The initial iteration of the thresher consisted of two major components - an outer barrel, designed as a net to prevent chaff from falling into the storage container, but allowing for seeds to fall through; and an inner abrasive drum, which would vibrate and rub the seeds along an inner abrasive surface, disturbing them enough to rupture their storage pods, and allow them to fall free of the stalk and the chaff. The rest of the parts are merely dedicated to bearing mounting, fastening, and motor powering. Similar to its drivetrain companion, this thresher was designed as a mere prototype, something to learn from, rather than a final product.
While I was not in charge of designing this portion of the project (it was primarily my partner), we learned a lot from this experience.
In regards to the actual design, we learned that the drum spacing was far too narrow, and that tolerances needed to increased. Secondly, we realized that the internal abrasive drum needed to have some way to eject stripped chaff, that would ideally come in the form of a horizontally oriented auger to drive exhausted stalk and raw biomaterial out of the front of the thresher. Finally, my partner needed to modify our bearing mounting points in order to create a standardized construction method that used standardized COTS parts, such as bolts, bearings, nuts, and standoffs.
In regards to developing design practices, myself and my partner learned to work with the same design standards, primarily adapting towards mine. This includes proper labeling of features and parts, proper organization of features and parts, and proper organization of files, assemblies, and part studios. Neither of us had worked on a project of this scale, or collaborated with another individual on an engineering project in this way (previously, I have worked with others, but they have all used my style of design organization). From this, we learned to streamline our workflow to work better in conjunction with each other.
As previously mentioned, the primary purpose of the header was to create a system capable of harvesting and intaking all of the raw biomass and biomaterial. For this system, I ideated a system involving funnels directing stalk a few inches off the ground into a row of rotating blades, which would slice the stalk. The stalk would then fall into an auger system, which would drive all of the raw biomass into a center intaking converyor system. The rotating blades would be a mechanically linked system, using a single geared up motor and a system of belts and chains to properly connect all of the power between different blade systems.
While we have not gotten the chance to test this prototype in the field, we have received some substantial feedback from the staff at Kuilima Farms. While our design is not the traditional system of oscillating blades, they said that it is potentially viable system, while not traditional. However, we would prefer to switch to creating a system which would involve oscillating blades, since I believe that trusting an industry standard for small scale harvesters is better than venturing with our own intuition.
In regards to our design, we realized that we should create better design practices in regards to greater vision - we should see the greater picture of the entire assembly before starting designs, even with smaller parts of the project such as this.
Iteration 2
Learning (partially) from our previous attempt. we made a few modifications to this upcoming iteration, which we restarted from the ground up due to a flaw in our design that would completely break everything with the changes that we intended to make.
First, we increased the module of our gears to increase the stability and rigidity, and made the full commitment to either outsourcing manufacturing of the gears, and/or exotic 3D-printing of the required parts. In this, we created a herringbone style design, which allowed our gears to boast self-alignment features, quieter operation, reduced backlasher, and greater tooth engagement (opposed to normal spur gears). Second, we changed our coupling systems and gearing systems to use COTS parts, specifically designing around motor mountings and proper usage of chain. Finally, we further focused our design of our internal gears for ease of assembly and disassembly, something that we should have adopted from the start.
Regarding the other flaws which we did not fully acknowledge, we further exacerbated the issue of manufacturing our gears out of sufficiently rigid materials, by committing to herringbone style gears, which would require either 5-axis milling or exotic 3D-printing. Additionally, we chose to stick with the differential gearbox system, but did reduce the complexity of the required gearboxes by utilizing chain in the connections.
One of the major new tools that we decided to use in the development of this prototype was the tool of FEA, or Finite Element Analysis, to evaluate the structural capabilities and failure points of our design. OnShape, the web-based CAD platform that we used to design our models, has a built in FEA, which we used to assess the material strengths and weaknesses (along with the potential deformation in our designs) of our designs.
This is an invaluable tool that will be used much more later in our design and iteration process.
It was after the completion of this iteration that we realized that using a differential gearbox system was a non-viable route. It provided very little benefit, for a lot of extra complexity and manufacturing cost, and using a differential steering system was a far simpler, cheaper, and more viable path to take. Realizing that it was time to change course, we decided to cut our costs and the time sunk into this design in favor of the more viable (and much more simple) differential steering system.
However, we still look many of the lessons that we learned from this experience with us. Moving forwards, we planned to avoid having custom complex parts (such as gears, sprockets, etc.), and use as many sparts created by FRC manufacturers as possible. FRC manufacturers were chosen because it eliminates much of the time and energy needed to produce these parts, and it also ensures reliability and stability in components.
With this in mind, moving past this point, we decided to standardize the equipment that we were using for the entirety of this project, primarily the bearings, motors, shafts, collars, gears, and chain (1/2" Hex Shaft, Bearings, Collars, Gears, and #25 Chain). This part standardization allows us to adjust our designs much more freely once parts arrive, as we will not be waiting on specialized parts to arrive (upon modification).
From this experience, we also learned the value of FEA; we can now accurately predict points of failure in our design, and understand when, where, and under what loads our parts will fail. With this newfound ability, we can better test and simulate structural loads in our designs to ensure rigidity, stability, and the desirable ruggedness.
Finally, I learned that in order to avoid expending unnecessary and extra effort, avoid pocketing the plates before the basic plate geometry is finalized - pocketing early has the tendency to break the assembly, part-studio, and render FEA useless; consequently, the pocketing has to be redone, along with all following features based off pocketing.
Similarly to the differential gearbox system, we learned a lot from our first attempt at building a thresher. Again, while I was not in charge of the designing and building of this part, we both learned a lot from the first iteration.
First, we increased the tolerances between the abrasive center and the outer cage, which allowed better movement and tolerances. Secondly, we realized that it would be more space efficient to scale the dimensions of the drum down, but in adding a vertical movement system in the form of an auger-like screw, this system would work equally as well. Third, we modified the side port ejection geometry to better catch the chaff. Finally, we added proper spacing and securing for the bearings and shafts.
For this prototype, we still were under the presumption that this would not be the final prototype, just a testing rig to understand the principles of which our thresher must operate upon.
As a secondary addition, this prototype was expanded upon to include all of the necessary operating parts, such as an auger, storage system, power integration mechanism, and proper coupling.
The first major lesson that we learned is that we, 100%, need an auger or some sort of vertical movement system. While it was previously ideated, it simply was not integrated into this round of design.
The next major improvement that needed to be made was in the former of creating a proper coupling shaft. Both in this iteration and the previous one, the shaft driving the rotation of the inner abrasive drum was not a proper shaft, and was rather designed as part of the drum. Quite simply, this would not work, as the high rotational speeds and potentially high torquing forces exerted on a 3D printed shift simply would not work.
Iteration 3
The third iteration of our drivetrain, the differential steering mechanism, was a far simpler method of drive that we choose to use. As previously stated, the differential gearbox system was simply far too complex, with far too little benefit, to be worth of prototyping, developing, and testing.
This new approach to our drivetrain mechanism takes many of the lessons that we learned in the past, and applies them to our new design. This design aims to create the most simple drivetrain possible, with the most benefit possible, avoiding custom made parts in favor of COTS parts. All of the gears, chains, sprockets (with the exception of drive sprockets), motors, and structural components are available COTS (Commercial-Off-The-Shelf), allowing for sturdiness and rigidity without risk of material failures. Nevertheless, the design also has a focus on redundant functionality, especially at points of failure (such as the chained power transfer from gearbox to output shaft); this aims to hopefully avoid material failures, and minimize the negative effects of such failures.
This is a very simple steering system, which runs both sides on independent motors, and the difference in rotations between the motors (directly) produces the steering that we need; this operates without the complexity of a primary and secondary gearbox(s) being linked through a differential gearbox, and eliminates many unnecessary points of failure.
A major part of this new iteration was rethinking our power draw. Previously (and I know this sounds really dumb), we weren't super conscious of the power draw that would be required of the drivetrain, thinking that as an alternative, we could always run our motors at lower RPMs to save power.
However, looking at efficiency curves, torque outputs, and the sheer current draw that would be required, we had to completely rethink our system of power. With the help of our advisor, Mr. Baffoe, we created spreadsheets to calculate the ideal amount of motors at a specific gear ratio to achieve the speed an acceleration we needed. These spreadsheets factored in the different amounts of friction, efficiency, energy loss, torque, and speed requirements to calculate the ideal amount of motors at the ideal efficiency.
In regards to the speed and acceleration, neither of these values were extremely high, and were in fact, extremely low, only around 0.25ft/sec, and 0.125ft/sec^2. Thus, since there was such low speed and acceleration requirements, and power draw was a significant issue with the previous iteration, we realized that we can reduce the amount of motors, and simply just gear them up. Since the only thing we really needed from our motors was high amounts of torque, we decided after our calculations that 2 CIM motors per side running at high efficiency, and geared 60:1, would be sufficient propulsion for our robot.
The introduction of an extremely high gear ratio gearbox was one of the most innovative parts of this design. For this design, we chose to package our gearbox as an extrusion from the main plate, connected by shafts, t-slots, and plates. This gearbox utilizes COTS gears, specifically from FRC, #25 chain (2:1 reduction), and Versa-planetary Gearboxes (30:1 reduction) to achieve our desired reduciton. However, as clearly shown below, this gearbox is pretty big and clunky, and could definitely work to have its size reduced.
While this design was a major improvement over the previous one, there were still many improvements that needed to be made. The first major improvement that needed to be made for the next iteration is the shrinking of the gearbox. The gearbox, simply, is far too big, and can be significantly reduced. The plan to achieve this was to angle the motors off-center, and collapse them closer together to create a more triangular profile (for which, 2 vertices would be motors, and 1 vertex would be the output). In addition to this change, we also contemplated switching the motor power train to the output shaft to sprockets instead of gears, but were uncertain due to the horizontal tension placed upon a cantilevered motor. Adding a cantilever support is a potentially viable solution to this, but in the end, adds a significant amount of effort or complexity to this system which is designed around simplicity.
We also learned some lessons about cost - simply put, it would cost too much to structure all 4 of our plates out of aluminum, and thus we turned to acrylic. Our outer plate is not being used for any structural loading, and thus does not require the same amount of strength that our inner plate does. For this reason, we chose to switch courses, and change out outer plate to acrylic, which will simply act as an outer deflection shield against any weather or debris - for this, 1/4" acrylic should be more than enough.
Regarding the overall chassis shape, however, this was very close to finalization, as I was closing in on a shape that was both functional, and aesthetically pleasing. One of the final changes that would need to be made was the pocketing of the inner plate to reduce weight, which I was not sure was necessary, as the inner plate only weighed around 15lbs.
The third iteration of the thresher design was a complete redesign. It incorporated completely different principles, such as continuous and counter rolling drums, which expedites the process of threshing, and distributes more concentrated loads. This iteration has been designed specifically to be larger to handle the load and the size of the plant.
One major design change in this iteration is the incorporation of a metal mesh. During pervious iterations (in which a metal mesh was not used), there was the potential from seeds not being fully separated from the plant itself (still attached to their pod and/or stalk). This iteration specifically addresses the shortcomings of the previous designs.
Finally, this iteration has an integrated storage system. This significantly reduces the internal complexity of the robot by not requiring more conveyorbelts to be integrated into the central internal design. This iteration also includes the proper mounting brackets which have been integrated into the greater chassis design using the standardized 2"x1" aluminum channel.
Iteration 4
From the previous iteration, we first learned that we needed to reduce our gearbox profile in order to increase and maximize space efficiency. To do this, I decided to angle the motors, moving them as close together as possible. Additionally, I reduced the tooth count on the gearbox gears, thus shrinking the distance between each individual motor and the output gearbox. These changes resulted in an angled gearbox shape, similar to an arrow. While it still used the same fundamental principles, the geometry of the gearbox has been optimized for space and compactness, further pushing it into the plates, while still saving room for the necessary wiring.
The creation of custom sprockets is also something that came along with this prototype. While these are not sprockets for the chain (we are using COTS sprockets), these are still equally as vital to the functionality of the final robot. These sprockets are designed to be the carriers of the bottom tracks. To prototype these, we plan to cut the sprockets out of 1/4" aluminum, and then use a central mounting bracket to connect the different sprockets.
Additionally, one of the issues that arose during this time of the project, was the electronics mounting. We needed to figure out a way to brace our electronics and systems against violent vibrations and impacts. However, upon examining this, we realized that it was equally as important for us to be able to implement this system throughout the bot, as violent shocks, impacts, and sudden changes in terrain could yield catastrophic results. To combat this, it was suggested that we design and implement a suspension system.
However, for this project, we had no idea what a potential suspension system could look like - we've never attempted anything of a similar nature, and it was thus difficult for us to foresee how to build such a mechanism. For inspiration, and design ideas, we turned to the designs of tanks, specifically looking at mechanisms such as a torsion bar suspension and sprung suspensions.
For this project, we chose to work with a torsion bar suspension, as our robot has enough width, and the torsion bar has a minimal profile, along with minimal complexity. Additionally, replacement of the torsion bar (upon mechanical failure) is relatively easy.
To work on our suspension and to ensure proper functionality, we conducted FEA to calculate the vertical movement that was available from such a suspension, which we found to be around 16.7 degrees, and with our 100mm lever arm, works out to be approximately 19mm of suspension. This is something that we are extremely happy with, and is just enough to be satisfactory.
Prototyping
Waterjet (Acrylic)
Laser Cutter (Acrylic)
Manual Mill (Sprockets)
HAAS Lathe (Shaft)
Bandsaw (Shaft)
Prusa MK4 & MK4S (Miscellaneous Thresher Parts)
Bambu Labs A1, Bambu Labs P1, Bambu Labs X1 (Various Chassis and Thresher Parts)
Sottering Iron (Heat Set Inserts)
Drill (Misc.)
Hammer (Misc.)
4'x8' Acrylic Sheet
4x 40:1 VersaPlanetary Gearboxes
4x CIM Motors
M8 Heat Set Inserts
M8 Bolts (Misc.)
M6 Heat Set Inserts
M6 Bolts (Misc.)
M6 Heat Set Inserts
M4 Bolts (Misc.)
M4 Heat Set Inserts
M4 Lock Nuts
PLA
PETG
One of the longest parts of the prototyping process was the manufacturing of our plates on the Waterjet. For this initial prototype, we wanted to use acrylic (as it was low cost and high rigidity). However, this meant that we would have to manufacture all parts in the shop. For the larger plates (main chassis structural plates) this requrired us to use the Waterjet.
This process was extremely time consuming and extremely irritating to do. The waterjet did not like working with the acrylic, as the low pressure constantly caused the garnet to jam. Additionally, the texture and material properties of acrylic meant that with every cut, there was a chance of the material shattering and creating sharp crystalline edges.
The Waterjet at Punahou has a habit of backing up and driving water into the garnet tube, which was another problem that we had to deal with during the prototyping phase of this project.
For the smaller plates in this project, we were fortunate enough to be able to use the laser cutter to cut out the acrylic that we needed. This was primarily for the thresher and the gearbox, and worked much better than the waterjet.
For this project, we needed to use 32 tooth #25 chain sprockets. However, we couldn't find any online that matched our required specifications (with a hex bore). We were able to find some 32t #25 sprockets in the shop, and with permission, we cleared out a center bore, and created an appropriate mounting bracket.
This process was done on the manual mill, and used the soldering iron for all of the required bolts (heat set inserts).
This project uses a lot of precisely cut shaft. This includes everything from the shafts used in the gearbox, to the chain shafts used in the threasher. To get the shafts to the precise length, we used the bandsaw to cut to an approximate length, and the lathe to turn the shaft faces down to a precise size.
One of the most helpful changes that was made during this project was the utilization of heat set inserts in almost every aspect of this project. These were extremely useful, as they allowed us to put high strength threads in 3D printed parts, allowing for quicker prototyping, and the avoidance of time consuming and expensive processes, such as buying metal parts and tapping them.
Chassis assembly was an extremely tedious process. While the chassis was not completely done, it had all of the important subsystems. Working with such large systems was extremely irritating, especially when the assembly was this complex.
One of the greatest challenges that we faced during assembly was the complex nature of the robot. With so many intricate parts in such a precise arrangement, it was like trying to put a puzzle together. The parts would have to go together in an extremely specific order, and oftentimes messing up this order would result in a partial or full dissassembly, and a complete restart to the assembly process.
Testing