Arduino For Small Farms

1/7/25 Research/Construction Update:

To date I have conducted research and construction in the Phase 1 areas of this grant: 1. automatic dehumidistat, 2. blueberry sorter, and 3. greenhouse reemay mover.  Below I summarize this progress:

1. Automatic dehumidistat.  In August I met with the participating farmer, Chuck Currie, Freedom Food Farm, and collected information on his current grain drying system and what improvements he proposes.  In September and October I built a test system to compare humidity sensors and collected data on temperature/humidity sensors that will be useful in March when I start construction of the bench-scale prototype.  I was able to use my clothes dryer as a proxy source of a humid airstream and this worked well, allowing me to passively collect data while doing the chores.  I am focusing on using the DHT-22 temperature/humidity sensor, as it is inexpensive and accurate enough.   Through analyzing the data from my clothes dryer, I think I now understand a technical property of the DHT-22 that was confounding my previous attempts to code the control logic on the differential humidistat; namely, that the temperature and relative humidity readings are out of phase by a matter of seconds, which makes using the data for determining absolute humidity a difficulty.  If true, this hypothesis would explain problems Anni Schmidt (who wrote the original code) described in our previous attempts at using these sensors on a grain dryer.  One remaining strategic question is whether or not it is possible to monitor the grain dryness directly, or whether, as I have assumed in the past, it is necessary to use moisture in the airstream as an indicator of drying activity.  This will be the first question I address in the Spring.  
2. Blueberry sorter.  In August I met with Dr. Mbanisi of Olin College of Engineering.  We agreed to prototype the blueberry sorter with a team of students from his Robotics and Mechatronics class at Olin.  In August we collected data on sorted blueberries (blueberries sorted by the farmer by hand) by taking digital photos of blueberries with a special shape and color recognition camera.  This data became the basis for training a computer to discern between the 3 proposed categories of blueberry: Ripe, underripe, and overripe.  The student team convened in September: Jacob, mechanical/electrical, Alana, electrical, and Luke and AJ, code.  Through the Fall I worked closely with Dr. Mbanisi and the team to develop a working prototype which will be used in field trials July 2025.  The blueberry sorter mechanically processes blueberries from a hopper to a single file of blueberries on a conveyor belt.  The blueberries pass under a camera that sends images to a microcomputer.  The microcomputer classifies the individual blueberries and then activates pneumatic valves that use bursts of air to separate the blueberries into 3 bins.  By the end of their semester, AJ, Alana, Jacob, and Luke had a benchtop prototype that worked in the lab (though in December it's difficult to obtain unsorted local blueberries to test the device).  The prototype sorts about a blueberry per second, has a materials cost of about $500, and an accuracy rate between 85-93% .  One other interesting feature is that it is really a "delicate object sorter", quickly adaptable to sorting any berry or seed.   There are still, however, mechanical and code issues that the students wish to address in the coming semester.  Field trials for the Sorter will begin in July 2025, and by then I hope to have two working prototypes so we can get the most data possible from the limited time during which we'll have fresh blueberries.  Adam and Carissa (the participating farmers) are ideal collaborators; they are available and supportive when consulted, and go out of their way to make the farm and blueberry sorting shack available to the engineers on the project.  I am cautiously optimistic that within the $500 for parts budget, a small-scale blueberry sorter can be made.IMG_6904 2.
   

   

   

   

   IMG_6778

3. Reemay Mover.  In August and September I prototyped a bench scale model of the Reemay Mover and wrote the software sketch and tested it.  By October I had a 1/4 scale model working in my front yard.  The design is based on a motorized roller for greenhouse side flap ventilation, a proven product already familiar to many winter production greenhouse farmers.  To date I've made three attempts to construct the Reemay Mover at full scale (50' x 30') in one of Roots Farm's winter production greenhouses.  The control, logic, and electrical (bits and wires) part of the machine works well.  The design principle of "fail-to-manual" works well in this machine; a farmer can quickly disconnect the motor and roll the reemay by hand.  So even without the "robotic" part of the device, it is still saving labor.  The mechanical part (the atoms) is proving difficult.  Problems of steering error correction, proper tensioning of the reemay, and tucking in the sides of the reemay have all been tricky.  I have poured work in to this part of the project and made some progress without drastically changing the design path or increasing the cost of materials in the machine.  However, it is also prudent to consider other design paths that use the same logic and controls but a different mechanical scheme.  The unforeseen mechanical problems with this project have quickly eaten up the planned labor budget.   However, because I feel like the problems can be solved, and I feel this design could have a large positive impact in New England, I'm committed to making it work.  I believe I can borrow labor hours from the other phase 1 projects and stay within the overall phase 1 budget.   Kelli and Mike (Roots Farm) are enthusiastic collaborators; their farm is already compact, carefully planned, and technologically advanced, making it a good fit for testing a technology like the Reemay Mover.