Arboroot

Inspiration

People want to be self-sustainable, but it's difficult. They want to cut their carbon emissions, but it's difficult. One main way that people can be self-sustainable is through producing their own food in their backyard, which is healthy, environmentally friendly, and affordable, but it's time consuming and laborious. I experienced this first hand when I watered my neighbors garden over the summer. I got so bored of standing with the hose watering their tomato plants, that two epiphanies struck me.

Gardening could be automated.
Even though I wanted to be self-sustainable and help the environment through gardening, I couldn't because it took too much time and was repetitive and boring as shown by my experience watering my neighbor's garden.

What it does

So, here we are today. These realizations evolved into Arboroot. People who want to be self-sustainable, healthy, and environmentally friendly through producing their own food in their backyard but are too busy to do so, can now spend just one hour per week on their garden. That one hour is the fun stuff too: harvesting, and putting little seeds in.

Our greenhouses come with a water tank, solar panels, automated drip irrigation system, and plant columns, controlled by AI machine learning and shown on the user’s website and app. Arboroot takes care of gardening for users through a machine learning model that is pre-loaded onto the greenhouse microcontroller, which “knows” (pulls from a plant API database) the amount of water to give to the plant. This water is delivered through our drip irrigation system (~60% more efficient than regular watering), which also includes a place for fertilizer. Our greenhouse plant columns allow for vertical plant growing, maximizing space.

All of the sensor information in the greenhouse including the humidity, temperature, soil moisture, water level, and water pumpage is sent to the user’s website which looks like a dashboard. We replicated all of this real life functionality through our proof-of-concept demo.

How we built it

Today, at Hack TJ, we focused on building up our software architecture with express.js and Arduino as a proof of concept, providing a basis for our work with real life materials. We learned how to connect hardware (gpio & arduino) to an express js server. We simulated the greenhouse hardware with real sensors and actuators; the frontend website is functional and can talk to the server to monitor and control the hardware; the server is robust (although not super secure) and serves as a baseline for future improvements.

Challenges we ran into

It was difficult using Websocket to connect with Director. We couldn’t find the correct route for Websocket to connect to, but with time, we figured it out! We also ran into the issue of Raspberry Pi overheating, so we had to switch back to an express server and arduino midway through our project.

Accomplishments that we're proud of

Successfully getting the sensor data to show up on our local website and updating to all the devices connected to the wifi network in real time. Talking to mentors and fine tuning our software architecture with lovely diagrams on the glass whiteboards Getting all of our devices (phones, laptops, etc.) communicating with each other on a local network - and updating live at the same time.

What we learned

In the future, instead of express.js and Arduino, we will combine both by building a Flask app that uses python files and GPIO pins (Raspberry Pi’s version of Arduino) to control the sensors directly. The Flask app will have the same functionality as our local website right now, but we will migrate to Flask simply because it uses Python, which makes it simpler to control the sensors directly and to integrate machine learning.

What's next for Arboroot

Our very next step is to do customer validation by asking a subset of the target market that is most likely to buy our product are vegan/vegetarian new homeowners and millennials who want to live sustainably and be healthy by growing their own food.