Project type: Group
Timeline: One semester

Our first year capstone project at Carnegie Mellon was sponsored by Microsoft. The brief was "designing for a world with a billion sensors." My two partners and I created FamilyPatterns, a software tool that takes advantage of the Internet of Things to help parents and children accomplish goals and understand behavior. Most of the semester was spent on research, so that we could understand our target user group and evaluate our designs in a futuristic context.

The final deliverable was a video explaining how the product is used.


The Project Timeline

Project overview

The video sketch:

How it works:

FamilyPatterns is a software program that recognizes families’ needs and attempts to solve complex problems by harnessing sensor data that’s already being collected by products inside the home. FamilyPatterns is a modular system, allowing users to pull in any data stream they want to, from medical data services, to wearable devices, smart appliances, and sensors from the outside world. The data are then anonymized and aggregated so that data scientists can constantly refine algorithms for the advice and problem-solving features. By collecting data constantly, FamilyPatterns can provide truly targeted recommendations to users.

Phase One: Territory Definition

In order to determine a problem space, our faculty advisers suggested we create a territory map to investigate our interests. My two partners and I kept returning to the idea of family relationships and intimate spaces; we felt the nuclear family would be an interesting space to design for because of the relatively permeable boundaries surrounding privacy. The opportunity to distribute wearable devices among members of a family was exciting to us as well. 

Phase Two: Exploratory Research

In-depth investigation of the problem space led us to choose dual-income families with toddlers as our specific user group. An online questionnaire, a cultural probe, a small experience sampling study (a "beeper study"), a literature review, competitive analysis, and expert interviews led us to that decision.

At first we thought that we could use sensors to increase the amount of time parents spend doing the things they love about parenting--even the serendipitous activities.

Our original design goals.

Data from our questionnaire showed that parents of toddlers experienced higher highs and lower lows than parents of older kids, when it comes to enjoyment of everyday parenting tasks:

We also learned from our research that dual-income families with young children felt overwhelmed by their lack of parenting experience, and that they welcome digital intervention that can help them feel like better parents.

Cultural probe materials were sent out to five families with children between the ages of 6 and 17. We chose families with older children so that the kids could answer questions we sent along, in addition to the mom and dad questions we sent. The respondents, from all walks of life, unanimously mentioned their feelings of "not having all the answers."

Phase Three: Generative Research

Our next step was to interview parents of young children and observe families undertaking common tasks in context. We visited several Pittsburgh-area families at their homes for perspective.

What we learned caused us to shift our design goals. Our interviewees had very different attitudes about documenting their children's milestones and health data, and photographing cute moments to share with extended family. But all the families we talked to described feeling like they weren't living up to the task of documenting their children's early years. These feelings superseded the design goals we identified in the previous phase of research. 

Our new design goals overlapped our previous ones:

To help define a set of features for our product, we created a persona based off a composite of some of our interviewees. To do so, we plotted our interviewees on a two by two matrix examining two important characteristics:

Phase Four: Design iteration and evaluative research

My partners and I jumped right into storyboarding after getting excited about our new design directions. We explored wearable technology and smart home concepts to help parents document and reflect on their experiences, as well as provide guidance for productivity and better habits--on the part of the child and the parent!

Taking these concepts back to our interviewees (and to a few fresh eyes) for speed dating was invaluable in helping us correct the ideas we had that conflicted with parents' other goals.

Speed dating with these first concepts helped us see what we were getting wrong. After this first round of testing, we felt frustrated with our handful of disjointed, lackluster concepts that were left behind. What helped us put things in focus going forward was realizing that if we were designing for a future world of a billion sensors, we needed a better concept of how the world of five, ten, or fifteen years from today would be different from 2014's world. So, we took a break from ideation and did some research on futurism, pinpointing a few future-world characteristics that ended up inspiring our final design:

With companies like Nest, Fitbit, 23andMe, and Kaggle already popularizing the smart home, wearable technology, the quantified self, and data science today, we felt that the predictions we were reading about the future of life and work weren't too farfetched.

All the products are connected. A global, immersive, invisible, ambient networked computing environment built through the continued proliferation of smart sensors, cameras, software, databases, and massive data centers in a world-spanning information fabric known as the Internet of Things.
— Pew Research, 'Digital Life in 2025'

Despite the changes in technology to come, we felt that the complex challenges that today’s families face regarding parenting and child development aren’t going to change anytime soon. Parents will always struggle to stay patient with unreasonable toddlers, or to make time to listen to their children and spouses.

Therefore, we sought to use machine learning and data science to harness the large quantities of data already being collected by families' belongings and environments. Maybe data science, appropriately packaged, could help families achieve personal goals as well as promote family togetherness.

Low fidelity prototyping

In order to gauge how users might feel about using the Internet of Things to support goals and family bonding, we staged several enactments with fresh participants who were new to our project.

We tested many futuristic hardware concepts intended to help families control the physical environment. However, after the tests, we chose to toss out the robotic dogs and surrogate neighbor concepts in order to focus on a purely software-based product. Our high-fidelity prototypes focused on figuring out what features were needed so that parents could interact with their data in their preferred way.

High-fidelity prototyping

At left: a sketch of the "results" screen provided when the user asks a question, such as "how can my family have a more peaceful, stress-free dinnertime?"  At right: options for home automation, giving the user privacy and control of the "smart home."

Preliminary concepts for data visualization, allowing the most pertinent information to rise to the top, with detailed charts and graphs available for those who want to understand the "whys" behind the recommendations.

Preliminary concepts for data visualization, allowing the most pertinent information to rise to the top, with detailed charts and graphs available for those who want to understand the "whys" behind the recommendations.

A sketch for a large display to help families gauge their progress toward goals, such as weight loss or learning about music. (We referred to this is the "refrigerator screen"--a large-ish display that could be hung up in a common space in the house, the way bulletin boards or calendars are in some households today.)

A first visual design iteration with the features we refined through testing.

I predicted that being sensitive to individual culture and personal identity would be an important factor for any digital product presuming to help users with personal goals. So, we tested a "social communities" feature with members of different identity groups: one group of devoutly religious parents from the same church, and another testing session with a couple of people from other identity groups such as vegetarianism.

A mock-up of how an individual user's identity groups would be acknowledged by the product.

Testing the "religious community" feature with parents.

Surprisingly (to me, at least) this concept completely failed! Our interviewees did not care to learn about how other members of their faith or identity community were accomplishing personal goals. Interviewees were also concerned about their personal data and history leaking into "social" spaces. We ended up excluding social factors from the product entirely.

We showed our high-fidelity wireframes to behavioral design experts as well as other people with young children over multiple rounds in order to understand exactly which features were necessary, and how parents prefer to receive information.

Phase Five: Visual design and finishing touches

We sought a visual design that would be appropriate for families, but clean and energizing as well.

Visual design iteration #1:

Visual design iteration #2:

Final visuals: