BRIEF

As part of a grad school project, our team was asked to design for a smart city using technology as a means to change human behavior. We addressed the topic of food waste by reexamining the way people use their food.


Context: MHCI+D Ideation Studio

TIMELINE: 11 weeks, September – December 2017

Teammates: Corey Brown, Lan Vu

KEY CONTRIBUTIONS: Research, user testing, hero flowS, video prototype, writing & editing
 

 

 

 
rye-hmw.png
 


 

The Solution

Rye helps to minimize food waste by taking into account your food inventory when making recipe recommendations. The incorporation of machine learning also allows Rye to learn the user's dietary preferences and grocery shopping behaviors when tailoring recipes.

Rye uses the Echo Show as its platform which is primarily designed for voice. Thus, you can use Rye by speaking to Amazon's 'Alexa'. Using the touchscreen is available too.


 

 
 
 
 
 


PROBLEM SETTING


"In wealthy countries, especially in the United States [...], around 40% of wasted food is thrown out by consumers."

"Globally, we throw out 1.3 billion tons of food a year, or a third of all the food we grow."

 

— NYTimes 'How Much Food Do We Waste? Probably More Than You Think'

 

 
 

Field Research

Observing behaviors

 

We took inspiration from The Little Free Library, a nonprofit organization with over 60,000 public bookshelves encouraging neighborhood book exchanges around the world, by stationing a shelf with a variety of foods in two different locations. We hoped this familiarity of community-wide sharing would resonate with residents and promote greater trust and participation in our experiment.

TAKEAWAYS

  • People expressed greater interest in giving away food versus taking food.
     
  • The focus of our neighborhood food pantry evolved from a food sustainability experiment to a sociological venture, as we learned that it wasn't so much about where the pantry was located as it was about the socioeconomic status of the people involved. 

 

ryephoto.png
 

 



Ideating & Down-Selection

Generating design concepts

 
  Pinned-up concepts with illustrations and descriptions

Pinned-up concepts with illustrations and descriptions



After consolidating data from our observations, we dove into ideating 30 potential concepts. These ranged from personalized apps to smart kitchens to community-wide efforts. We used a combination of methods like mind mapping, 2x2 frameworks, and 8x8 sketches to brainstorm. 

 

 
 
 

Once we generated a handful of ideas, we narrowed down to five by evaluating how needed, feasible, and inviting their potential was. We were deliberate in selecting a diverse group of ideas, so as to make sure we didn't leave out any considerations. From there, we drew out storyboards to help visually communicate our designs. 

We presented them to the class during a pin-up critique and got feedback that helped us to eliminate two ideas. Comments around technological capabilities and similarities to an already-existing product that would have been hard to differentiate from helped us to solidify our final three concepts.

  A storyboard concept I drew for a mobile app that uses scanning technology to detect when food has gone bad

A storyboard concept I drew for a mobile app that uses scanning technology to detect when food has gone bad

 

 

 


 

Prototyping

Getting feedback from participants

We were asked to build low-fidelity paper prototypes for our three concepts we narrowed down to—this particular approach was useful in allowing us to quickly develop functional representations of our design ideas to test out the flow and experience. We wanted to answer the following objectives through user testing:
 

/ Understand what people are currently doing with their unused foods. Do they wish there was a better way of disposing said foods?

/ Find out if each prototype effectively addressed the issue of remedying food waste at the consumer level.
 

 
prototype2.JPG


Prototype 1 — FoodMatch App

A mobile app encouraging neighbors to give, receive, or exchange food.

"This kind of reminds me of Craigslist...are [the people] safe to meet? I give away a lot of clothing...reminds me of that but with food." – Participant 2

Cons: Delivery feature was unclear, as well as who should be responsible for delivering the food.

 
 
 
prototype1.JPG

Prototype 2 / Meal Prep App

A mobile app with recipes that optimize food in your inventory.

"It's a little more of an everyday thing. I can just use this if I need to whip up something quick." – Participant 5

Cons: Not obvious that recipes were based off of the users' food inventory, and were in order from most to least efficient use of food.

 
 
 


Prototype 3 / Meal Prep Assistant

A voice + graphical user interface that makes optimized recipe suggestions.

"Being able to control it without touching is really great for me." – Participant 3

Cons: Wished the system was "smarter" and knew the user's schedule and food preferences.

 

 
 

Insights

What mattered most to people

 

People felt bad about wasting food and were open to an alternative outside of composting. They preferred to make more efficient use of their food before it spoiled.

Time was a determining factor in whether or not participants would engage in alternative efforts to reduce food waste.

Underlying factors, such as security and level of effort, impacted participation.

 


We initially wanted to address food waste through a consumer sharing network, but our field research from and user testing proved that this idea did not fit with the consumer’s mental model.

User testing revealed the most potential for an application on the Amazon Echo Show, for participants found the experience of using a VUI enjoyable, innovative, and practical, especially the hands-free aspect while cooking. Thus, we iterated on our third paper prototype to develop Rye. 

 

Use Cases

How Rye works

To best illustrate the user's journey through Rye, we developed three likely scenarios:

 

Onboarding

We chose to develop an onboarding flow to familiarize a first-time user through key features such as the home screen, sidebar navigation, inventory, and recipe steps.

onboarding.gif
 
 
 

Pre-Recipe
(with low-confidence)

We chose to develop the pre-recipe flow in the event that Rye has “low confidence” in an ingredient used for a specific recipe. “Low confidence” means that Rye in unsure about the status of an ingredient, for instance, if it was purchased a while ago, has never been used, or if it's expired. 

 

lowconfidence.gif
 
 
 

Recipe + feedback
(with ingredient substitution) 

We chose to develop the recipe + feedback flow in the event that a user does not have an ingredient after starting a recipe. Rye makes ingredient substitution suggestions based on your inventory or suggests adding it to your shopping list.
 

 

substitution.gif
 
 
 

Lessons Learned


Interactions flows for Rye were tricky, as our team often ended up in confusion or disagreements around it. Looking back, I wish we had placed more emphasis on developing the flows earlier in the process and committing to the decisions we made. Remember that time is limited! Sometimes, you just have to make a call and stick to it.

Towards the end of our design journey, we visited an Amazon bookstore to play with the Echo Show. The experience was an eye-opener. We had to go backwards and make last-minute visual changes on our UI after realizing just how small the screen was. There were also limitations in capabilities and syntax when speaking to Alexa. We felt much more informed after this visit and decided to include a Voice Design Guide section in our UI Spec to help users get over the learning curve of using a voice-first device. Always make sure you physically interact with the  device you're designing for. It's not enough to rely on videos and articles of other people using it.

 
 

Future Considerations
 

 

1. Explore ways to capture current inventory for items that have already been opened and/or partially used.

We made a conscious decision to leave out this feature in Rye, simply because our original goal was to have minimal effort on the user's part. The task of remembering and measuring opened food and then reporting that to Rye would have put far too much responsibility on the user. 


3. Create built-in timers to help the user during the recipe process.

We received multiple comments about incorporating timers to help ease the cognitive overload. Cooking isn't straightforward, so setting timers for each step could be very useful in creating a more streamlined experience.

 

 

2. Account for foods eaten or used in recipes outside of Rye.

We know that people won't use Rye for everything. Finding a way to incorporate the consumption of foods not used in Rye would make inventory calculations more accurate.
 




4. Expand Rye outside of Amazon and Whole Foods.

People buy food from a variety of locations. If we could develop partnerships with other supermarket brands, Rye could become more inclusive of the user's entire food inventory rather than just a portion of it.