Green Thumb
Personal | June 2020
Overview
The Green Thumb project was an exploration into automated plant care and maintenance. Over the course of the project, I investigated many different approaches to solving generally the same problem: how to efficiently and consistency water house plants. As many house-plant-owners have experienced, keeping plants alive can be a challenge especially for those of us that weren't blessed with green thumbs. The following prototypes showcase the different strategies I took to address this core problem.
V1 Prototype
My first prototype addressed the watering consistency problem with a Raspberry Pi Zero and water sensor. I designed and 3D printed a housing to contain the wiring and to allow the device to penetrate the soil of a house plant. I then wrote a python script that would send a text message to a predefined phone number indicating it was time to water the plant. If no water was detected after a period of time, it would continue sending messages with increasing frequency. However, once the water sensor detected that the plant had been watering a timer would reset and go dormant until it was time to water the plant again and the cycle would repeat. All in all, this prototype was moderately effective, with the main shortcoming being the reliance on the user to still facilitate the watering.
V2 Prototype
Version 2 of the Green Thumb project transitioned away from a simple reminder with validation, to an automated water distribution format. This prototype consisted of a microcontroller, a solenoid valve and hosing, a reservoir that was fitted with small metal stakes to secure it into the soil, and 3D printed electronics housing to be mounted to the outside of the pot. I programmed the microcontroller at first to simply open the solenoid valve for a short time on a predetermined schedule. I later also implemented a simple API endpoint that would allow the solenoid to be triggered over my LAN. Overall, this prototype worked very well, the user could set up the system once and could forget about it for 4-5 watering cycles. However, the user would then need to refill the reservoir. This solution would be just fine for a single plant, but I wanted a solution that could maintain a number of different plants at once and this system didn't scale very well.
V3 Prototype
The next iteration of the Green Thumb project consisted of a single large reservoir, a microcontroller, a scalable number of peristaltic pumps, and lengths of thin tubing. I programmed the microcontroller so that each pump could be triggered independently via a predetermined schedule or via API over my LAN. Since each pump could be triggered independently, they could be configured to meet the varying water demands for different species of plants. For instance, one plant was a able to receive small amounts of water twice a day, while others could receive a larger amount once per week. The single reservoir reduced the need for refilling to only a few times per year. This prototype, while wildly effective at reducing the user oversight needed to keep plants alive, did have a significant flaw in that it required lengths of tubing to be run across the living space. In the event the user had many plants distributed across their living area, the tubing could get unsightly rather quickly.
V4 Prototype
The final iteration of the Green Thumb project was a reframe back to the basic requirements of the project, it reemphasized simplicity, aesthetics, and ease of use. Instead of relying on electronics to facilitate plant watering, this version relied on a reservoir beneath the pot that wicked water into the soil via capillary action. A user could keep the plant potted, insert the plunger with the wick attached into the soil from beneath the planter, and slide the planter into the sleeve. This version accomplished being able to provide individualized autonomous watering to different types of plants, without the need for tubing or a power source.