Social Patch

Academic Research submitted for the Conference
FUSE DIS 2016 - Designing Interactive Systems

Key Deliverables I produced for this Project:

  • Ethnographic Research
  • Physical Prototypes
  • Feedback Surveys
  • User Testing
  • Research Paper
  • Duration 3 months (2016)

    Project Overview

    Some screens of the Form Builder project

    How did I end up in this project?

    For my last course at The University of Queensland, I took Special Topics of Design Computing which was a great way of rounding up my two year program. The main core of this course was to explore connected realities using augmented reality or the Internet of Things (IoT). I decided to go with IoT because adding sensors to objects or ambients seemed an amazing idea considering only in limited courses I've had the chance to explore this kind of technology.

    We all pitched our ideas to make teams. I wanted to explore how to enhance a sense of belonging to a community through a connected shared garden (all this based on the Internet of Things technology).

    The project iterated a lot during the summer semester, so I will guide you a little bit around which changes were done, which were the most outstanding ones and the rationale behind them.

    I'll be dropping some comments regarding the process around this area

    The methods used to inform the early stages of this project were the following:

    • Academic papers, Official Gov Reports
    • Surveys
      Targeted to specific community garden members
    • Interviews
      To community garden members and gardeners


    Ramya, Krista (my teammates) and I started reviewing published information on what other community gardens around Australia were doing to foster cohesion and explored the dynamics of Community Gardens (CGs) around Brisbane. We then joined the Facebook groups of some of these CGs to gain deeper understanding of their regular dynamics, roles and motivations. Later on released a kickstarter survey with the CG members of these Facebook groups which allowed us to understand a bit more the context.

    Screenshot of Google Survey
    A screenshot of the Survey we released to some members of three community gardens in Brisbane

    In this survey, we gain insights about their motivations and their frustrations when interacting in a CG, how engaged they were, what would they change to improve the conditions of their CG community and how would they envision technology helping them out in this environment. After reviewing the challenges a CG presents (such as administrative, fundraising, networking), we decided to reduce our scope and focus only on how to promote human interaction between members through the garden itself to generate a sense of community.

    We envisioned that in this setting all ranges of the population can interact freely having a common goal: keeping the community garden alive and thriving. Kids could develop respect for nature through this dynamic and become more engaged with sustainability concepts from early ages. Elderlies could feel they are part of a meaningful group and improve their quality of life. Young adults that wanted to have a healthier lifestyle could learn gardening skills from the experienced members. This way, neighbors develop a sense of belongingness to an area and contribute actively in improving the community gardens and the social cohesion in the neighborhood.

    We also jumped back to the literature to find out what had been published regarding community cohesion through technology, sensors and their impact on human-behaviour because we thought those core concepts could inform our project better.

    The Idea

    In this stage, we decided we would have two specific settings:

    Diagram of The Idea
    Diagram of the idea: 1) Sensors collect data from the plants 2) The data gets send to the Home Device
    and displays a feedback 3) The members of the community garden react and interact with the garden

    In the Community Garden:
    Using a Photon (IoT hardware that integrates different sensors), we would be retrieving soil moisture, temperature and humidity data through sensors installed in what we called The Garden Device (In the Research Paper, we refer to it as the Rooftop Device). This Garden Device was located in a patch of soil with plants, and allowed us to detect when the plant was receiving too much (or not enough water), the humidity and temperature around this living element.

    Photon by Spark
    Image of a Photon by Particle (Spark)

    At Home:
    Each member of the Community Garden would have a Home Device located in their kitchen, living room or pretty much anywhere they wanted that had wiFi connection. This Home Device (also powered by a Photon) had a Neopixel (addressable LEDs) that lighted in a different colour depending on the readings collected from the Garden Device.

    Neopixel LED
    Image of the Neopixel we used

    So let’s say, the Garden Device detected that the soil was really dry today so it triggered a red light in the Home Device. This way, the members of the community garden would then know that the plant needed water and would go to water the plant
    … or so we thought.

    Neopixel and sensors connected
    Image of the prototype for our first iteration


    Building the System (The Backend)

    From the Garden Device we were retrieving data 5 minutes. We generated webhooks to retrieve the values and stored them in our database, collecting them by ID, type, reading, datetime, eventID and location.
    To make our Home Device work we were performing a query on our database, so if let’s say, the last 10 readings retrieved from today were in a range considered dry soil we would trigger a red light in the device. And so on with each sensor. This stage meant a lot of calibration and defining ranges we could use for each sensor so we could run some tests with participants.

    Simplified diagram of the solution
    Simplified diagram: sensor collect data from plant, it gets sent through the Photon by Particle (Spark) to the cloud
    First Iteration
    Ramya and Mayi
    Ramya and I, preparing the first demo with users

    We took our devices to a Community Garden that we placed under a lab setting.
    We invited elementary and high school teachers to interact with the system so we can get major insights about our devices.
    We got great ideas (mostly related with making it for school environments and gamifying the experience) but there was one that gave us a new perspective:

    quote icon
    I would like to know when the other members of the garden are around, to join them
    quote icon
    Ramya and Mayi
    Photo of the Community Garden setup, located at UQ's General Purpose South (Building 78)

    With this new idea, many hidden motivations arose:

    quote icon
    I am not very skilled in gardening. Sometimes I want to ask others if I am doing it right or to get some advice regarding how to take care of some plants I am not very familiar with
    quote icon
    quote icon
    Sometimes you just want to have a chat with people that have a common interest with you. I enjoy gardening a bit more when I have company. It helps me feel less lonely
    quote icon

    For most of the people we listened to, the community garden was directly linked with sharing with others, so it became very evident that we needed to explore how to make this space more social through technology. We weren’t quite there yet with what we had so far.

    We were questioned as well on how the system would handle scenarios where the garden is self sufficient:

    quote icon
    What if all week the weather was great, the temperature was low and you had a bit of rain during the nights? People would stop going to check the CG because they can see in their Home Devices that the plants are ok and have been receiving enough water. The social factor would be gone.
    quote icon
    Second Iteration
    Prototype including the Passive Infrared sensor
    Prototype what already includes the Passive Infrared sensor

    So if the social aspect was going to be a major part of it, we needed to figure out a way to encourage with the help of our little system the CG members to want to go and check their patch.
    We added a Passive Infrared Sensor to detect when someone was around the community garden. A passive infrared just detects events, so every time someone walked next to the Garden Device, we would count it as one event and we pushed that to the database.

    Prototypes ready for the next iteration
    Prototypes ready. Left, the Garden device. Right, two Home devices

    Every time that this presence event was detected, a pink light would be triggered for a couple of seconds, because the garden was being attended, or loved. The pink light would only be active while the person was around and detected by the sensors of the Community Garden, as a way of representing that someone was in the garden.

    Neopixel and sensors connected
    Final diagram of what each colour displayed in the Home Device meant

    We tested it again, allocating a Garden Device in the lab community Garden and then we gave the Home Device to some participants during a couple of hours. We explained to all participants what each colour meant. We asked them to do their regular things (study, do homework, work, having lunch) while they should try to monitor what was going on in the device. While the participants have their Home Devices, we would be triggering the pink light on purpose to monitor their reactions and capture their comments.

    The first insight collected was that very few of our participants detected when we triggered the pink light. They kept on doing their regular activities, so when the pink light was triggered they didn’t notice the change. Therefore very few of them reacted significantly. A very low number of participants decided to attend and check on the plants when they saw the pink light being triggered.

    quote icon
    I only came to see who was around because I was curious of what triggered that pink light
    quote icon
    quote icon
    I didn’t notice the pink light because I was really busy doing my homework but I wouldn’t want it to have a sound alert. A sound alert would make it very intrusive and I would actually end up annoyed
    quote icon

    We found that a lot of users missed the fact that someone else was in the garden “at that specific time”. Our participants stated that if they were busy (watching TV, cooking, working online) it would be very hard for them to notice a change of light at all.

    Third Iteration

    In the last iteration, two of our five participants mentioned that if plants were given feelings or an abstract representation of them, it could probably generate a better reaction that just identifying states present in the garden

    Instead of Dry Soil: Thirsty Plant
    Instead of Unattended Plant: Lonely Plant

    Working again with the Passive Infrared, we decided to make something a bit different. We showed the pink light feedback when someone was detected, but aside from this we triggered a cyan light when the community garden has been lonely for a day. We detected there wasn’t a lot of engagement from participants to react at all when another person was around the garden… now we wanted to explore how members of the CG would react when they realised the garden has been unattended, or was feeling lonely.

    Screenshot with part of the code that shows the passive infrared trigger
    Screenshot of the triggers we used when retrieveing data from the sensors.
    The Lonely value was added in this iteration.

    In the database, we would do a quick query to retrieve if someone has been around the garden in the last 24hrs. If the value was zero, we would trigger a change of light in the Neopixel located in the Home Device, so members would know that no one had visited the CG that day… or that the plants were feeling lonely. We expected that would be interpreted as an invitation for members of the community to go and visit them.

    Screenshot with Spark and Facebook logos
    Screenshot of the triggers we used in IFTTT using the Spark + Facebook channels.

    We integrated as well the social media factor during this iteration. We created a Facebook Group for the Community Garden and invited the participants to join. The idea would be that if the users missed the Home Device feedback, they would be able to see it in their Facebook feed. What type of message did we publish in their Facebook feed you may ask? We used the IFTTT platform. This platform allows you to create a trigger using different channels. Using the Spark and Facebook channels we created a trigger, so every time a change of status in the community garden was detected we published a message in the Facebook Group that would appear in the user’s Facebook feed.

    Screenshot with Spark and Facebook logos
    Screenshot of a Facebook status triggered with the IFTTT integration

    So, if you were a Friend in Facebook of the CG, you would see an image post that stated the Mint was very thirsty, inviting you to give them a bit of water. This was also an experiment to track if using mascots or personified plants generated a stronger connection with users of the system, as previous participants during the second iteration stated.

    Screenshot of some of the interactions in FacebookScreenshot of some of the interactions in Facebook
    Evidence of some of the interactions captured in the Facebook Community Garden group

    Even though not all the participants interacted as much as we would have wanted in the Facebook group, we did have a few that used the group in real time. They used it to figure out what was going on with the community garden as they detected some changes in the plant’s status and wanted to know if anyone had some answers about that.

    Feedback collected from staff and students around the campus that were part of the research

    Feedback & Improvements

    Here are some of the lessons learned:

    Some participants did react to the Cyan light and attended the community garden. One participant stated that the fact that he knew that the garden was feeling lonely made him try to comfort the plant.

    Some participants waited a couple of minutes after they detected the Cyan light to see if someone else would attend to the garden first”.

    quote icon
    I was really busy when the solitude light went on. I waited to see if some one else could attend first. I would have gone as soon as I finished my chores, but luckily some one else was there pretty quickly.
    quote icon
    Image of the final Home device prototype
    Final prototypes of the Garden Device (left) and Home Device (right)

    The 'not so' good things

    In this specific iteration, we detected more insights related with privacy and surveillance perceptions (which we didn’t detected in the earlier iterations)

    Most of the participants stated they felt the system was becoming a bit more intrusive because they were not sure what the system was monitoring from their input.

    quote icon
    Is it monitoring me passing by or is it actually detecting who I am?
    quote icon

    A participant that was really into gardening stated:

    quote icon
    This is not really accurate system, you know? Some plants don’t need that much water, but if the sensor detected that the soil was dry people would be overwatering it. The sensors need to be calibrated with the help of people that actually know about plants. They should be able to be customised (the sensor readings) depending on the type of plant, the type of soil and the amount of shade they receive.
    quote icon
    Image of the final garden device prototype
    Final prototype of the Garden Device

    Participants stated that not having a scale of measurement visible (to know how much water the plant needed and how the system was detecting the output) created a bit of discomfort, specially for participants that were not experienced in gardening and would rely solely on the Garden device feedback. This is related to previous literature we found, where not having a clear feedback to define a level of expectation to users can generate a bigger surveillance perception.

    Once we stated to all participants that their identity wasn’t being tracked, they stated that they felt it less disturbing.

    quote icon
    I would like to know which neighbors are the ones in the garden in case there are ones I want to avoid… but I don't want the garden to identify myself.
    quote icon

    One of the participants stated, acknowledging that identity tracking in IoT devices can generate intrusive perceptions.

    This project was exhibited as part of the Provocations and Works-in-Progress inside the Conference Designing Interactive Financial Systems DIS 2016. The complementary research paper was submitted as part of this delivery.

    You can find our research paper in the ACM Digital Library:
    Cervantes, Maria, Ramani, Ramya, Worthy, Peter, Weigel, Jason, Viller, Stephen A. and Matthews, Benjamin Robert (2016). Could the Inherent Nature of the Internet of Things Inhibit Person-to-Person Connection? DIS '16 Companion: Proceedings of the 2016 ACM Conference Companion Publication on Designing Interactive Systems. June 2016. Pages 177–180.
    ISBN 9781450343152


    Favourite Work

    Check out my other projects too