What it does
Project Sunflower is a proof of concept for a tracking solar panel device equipped with various other sensors connected to a MongoDB database for analysis in order to determine
How we built it
We let the engineer have free reign.
Challenges we ran into
Attempts to utilize a DragonBoard or a Raspberry Pi as the primary computer rather than a laptop, something that went through quite a lot of attempts that all ultimately failed such as attempted connection through ethernet for SSH on a Pi, attempted to use FastBoot to switch the DragonBoard to Linux, another try with the DragonBoard with an attempt to flash Linux with a microSD card using an Android phone as our SD card read/writer, connecting the DragonBoard to computer peripherals and trying to run Arduino and MongoDB libraries through Android, but these all ended up not working, with the last one being the closest attempt.
Accomplishments that we're proud of
We only broke one Raspberry Pi, a motor, and many many many cables through soldering and splicing to various sensors. Managing to execute our initial plan was satisfying in the sense that the transition from brainstorming ideas to blueprinting and referring to mentors to physically building and coding the hardware and software functions with positive mentor support and advice felt great in maintaining our drive.
What we learned
DragonBoard being in Android is quite frustrating and all of our challenges would have been solved with the correct IO connectors. For all of us this was our first experience working with a database software such as MongoDB, which was less painful to connect with out equipment than anticipated and pretty quick to utilize. Our main issue is the lack of options with MongoDB's data visualization options, although in our limited time we managed to make do quite nicely, with the complaints mainly deriving from knowledge of more dedicated visualization software such as Tableau.
What's next for Project Sunflower
Trail run scalability of measurements from multiple sensor clusters and organize these clusters using geolocation to determine best locations for plant placement
Connect MongoDB to better data visualization software such as Tableau through its paid version and extensions.
Once multiple clusters are set up, place them on smaller rovers for movability across and landscape rather than placement.