Orion: Your Sustainability Teacher 



How might we teach people to comprehend the environmental impact of their daily home behavior and foster lasting behavior change to improve human sustainability?


The Orion system is an in-home IoT device that monitors 4 utilities (energy, gas, water, and air quality) and relates those utilities to local sources to promote sustainable behavior change. In addition, it matches users to environmental organizations based on the user engagement patterns.  


16 weeks

My Role

Research, Educational Integration, Concept Development, and Product Design.


Mary Tsai, Jesse Wilson, Adrian Galvin, Laura Rodriguez


The Orion System

Orion moves the user through a sequence that is personalized based on the interaction patterns that occur between Orion and the user. The base sequence is as follows:

  1. Orion monitors 4 utilities (energy, air quality, gas, water) and provides information and suggestions to maintain a more efficient home.

  2. Then Orion begins to relate the user's utilities to their sources (e.g water to the water source) to promote a connection between the user and their surrounding environment.

  3. Finally based on the user's engagement, Orion connects the user to environmental organizations that they may be interested in.

Orion uses multiple levels of engagement to provide the user with information and promote behavior change. These include ambient visualizations, projected visualizations, and auditory interactions, as well as a variety of motivator tactics including light, sound, and movement.  

orion system.png

Ambient + Projected Visualizations

The first level of information is provided through ambient visualizations, which provides the user with an overall sense of how their utilities are performing in the moment. They use visual metaphors to represent the different utilities that are being monitored. This provides a mental model for the user, as well as an intriguing element that will help motivate the user to interact with Orion. 

The second level of information is provided through projected visualizations, which provides the user with detailed information about each individual utility. The user is able to see information about their usage patterns and their goal progress on a daily, monthly or yearly time frame. In addition, this is where the user will be able to access information about the sources of the utilities, as well as information about the recommended environmental organizations. 


Orion App

The Orion app allows the user to remotely access realtime utility data providing the flexibility to keep track of information while on the go. In addition, the user is able to see an overview of past and current goals + success rates, as well as create new goals on the app. The user would also be able to customize their notifications and privacy preferences through the app. 


How I Got Here?


01 Defining the Problem Space

The project brief focused on the following question:

How can artificial intelligence be utilized to support education and learning?

Based on this question, we began by defining our common areas of interests and the different stakeholders that would be included in these areas. We focused on volunteering and environmental advocacy as our high-level topic of investigation because we believed there was opportunity to use AI to help individuals be more informed about how their choices affect the environment. 


02 Understanding the Problem Space

We conducted multiple rounds of research throughout the project focusing on three phases:

  1. Exploratory Research, which focused on understanding the topic area we had identified

  2. Generative Research, which focused on understanding the user and user painpoints

  3. Evaluative Research, which focused on validating concept features

Each research phase informed the development of both the concept development and subsequent research phase.


03 Exploratory Research

During the exploratory research phase, we performed literature reviews and expert interviews to gain an understanding of the fields of learning and education, volunteering, and environmental advocacy.


Literature reviews

We performed literature reviews focusing on the topics of learning and AI in education. By understanding how AI is currently being utilized in the education field, it provided us with a base to build upon. 

In addition, understanding the methods of teaching and learning allowed us to develop a structured learning method for our final design.


Regardless of type, people want to have purpose and be able to do something with what they are learning.

Expert Interviews

We conducted 9 interviews with individuals who worked at environmental organizations and academic experts in the fields of social sciences focusing on the following areas of inquiry:

  1. How can citizens develop a personal connection to environmental issues?

  2. How might volunteers gain confidence and empowerment through the act of advocacy?

  3. How might Artificial Intelligence improve communication between volunteers and advocacy groups?

ACES volunteers have to learn and grow into their new advocacy role.
— Employee at environmental organization
By the time we hear the details it’s sometime too late to form a coalition. If we could come together more quickly that would help.
— Volunteer for environmental organization
Expanding a person’s frame of reference for where their responsibility begins and ends.
— Academic expert in sustainability and design

Key Insights: Exploratory

After the exploratory research phase, we collected the following key insights.

  1. People take action when they become invested in a narrative or future vision.

  2. Environmental campaigns must help their members grow and learn new competencies.

  3. Volunteer commitment and scheduling are key breakdown points.

  4. Organizations lack a consistent way to identify and communicate with potential volunteers.

  5. AI could be used in large-scale coordination and collaboration through convenient gateways allowing access for a range of individuals.

Design Principles: Exploratory

We developed the start of our design principles based on our initial round of key insights. After each phase of research, we revised our design principles based on our new key insights. 

  1. Our design should facilitate connections between community members and issues they care about.

  2. Our design should empower community members to take concrete action.

  3. Our design should help organizations coordinate collective action.

  4. Our design should have clear learning objectives focused on action-based learning outcomes.


04 Generative Research

During the generative research phase, we conducted a survey, digital diary study, and participatory design workshop. 

During this phase, I took the lead on developing the research strategy plan. I lead the development of the survey, digital diary study, and workshop timeline and activities, in addition to research synthesis. 


Survey: Volunteering

For the survey, we focused on gaining an understanding of people's experiences with volunteering and the reasons why they may or may not participate in organized volunteering. We received 40+ survey responses.

We found that time and the lack of concrete results played major factors in people not participating or feeling dissatisfied with volunteering.

Screen Shot 2018-08-10 at 10.51.25 PM.png
Wonder whether volunteer activities make enough of a difference to justify time spent.
— Survey participant 10
I’m more likely to do something if I can learn, be part of a commuity, or meet people with similar values.
— Survey participant 25
It’s good to help those outside my community but there is additional value to providing for family and friends.
— Survey participant 30

Digital Diary Study

We performed a digital diary study with 11 participants gaining insights from 6 locations, including New York, Florida, Australia, and Canada. It focused on gaining an understanding of how often awareness of environmental issues came up during a typical day.

In addition, to the daily questions we sent, we asked participants to take one photo a day of something they felt represented an environmental issue in their life.

Screen Shot 2018-09-02 at 4.14.07 PM.png

We observed that a few participants began to exhibit behavior changes throughout the study, specifically with the list of environmental activities we provided each day. At the end of the digital diary study, we asked each participant if their activities or awareness changed over the week and the majority said that they were more aware of the small actions they could be incorporating into their daily lives. It demonstrated the power that awareness plays in driving behavior change within people.  

Just being a bit more aware of my actions - though I haven’t changed behaviors, I have watched more closely.
— Participant 1
Yes, your list of activities made me consider how easy it is to undertake small tasks on a daily basis.
— Participant 5


We conducted the participatory design workshop with 9 participants. Working in teams of 3, they were tasked with performing a sequence of activities that would build on one another. 


Workshop Activities

  1. Personas: Develop two personas (one that cared about the environment and one that was indifferent about the environment). We provided personas templates that focused on identifying life experiences and current goals/desires that influenced their personas' current views on the environment.

 Building personas

2. Timeline (Part 1): Map out the life journey of each persona focusing on the events that influenced their current views on volunteering and environmental advocacy. We provided a storyline toolkit that included a timeline, visual and verbal elements to facilitate expression.


3. Magic Device: Design a "magic device" that would make connecting to environmental issues easier for the personas they developed. We provided a toolkit that included velcro modeling pieces so participants could quickly prototype their devices.  


4. Timeline (Part 2): Adjust the life journey based on how the "magic device" changes the perspectives, actions, and futures of the two personas. Participants built on top of the original timelines they created using a different color and storytelling toolkit.


Key Insights: Generative

After the generative research phase, we ended with the following key insights:

  1. People feel that their contribution does not make a difference because they cannot see visible concrete results.

  2. Time is major painpoint for participating in volunteering especially for young professionals.

  3. People are aware of environmental issues but do not actively engage in environmental talk or actions if it does not actively impact and improve their day-to-day routine.

  4. There is a push and pull between focusing on personal values (e.g. family and friends) and advocacy values (e.g supporting causes).

Revised Design Principles

Based on the insights gained during the generative phase of research, plus our initial design principles from the exploratory research phase, we developed revised design principles.

  1. Our design should focus on small scale, daily life activities.

  2. Our design should make the impact visible.

  3. Our design should create lasting social connections.

  4. Our design should utilize awareness as an empowering catalyst for behavior change.

  5. Our design should connect personal values to a larger environmental context.


03 Concept Development + Evaluative Research

Using the design principles we established, we developed 3 high-level concepts. These concepts focused on contextualizing the user's actions on the environment within a home setting through three different medias: Data Visualizations, Virtual Reality, and Voice User Interface.

  Concept 1: Invisible to Visible:  Waste and consumption are visualized by an in-home device

Concept 1: Invisible to Visible: Waste and consumption are visualized by an in-home device

  Concept 2: Immersive Experience:  This system shows a user what their environment might look like if different choices are made

Concept 2: Immersive Experience: This system shows a user what their environment might look like if different choices are made

  Concept 3: AI Home Agent:  A contextually aware AI agent that engages the user in conversation during their daily routines and practices

Concept 3: AI Home Agent: A contextually aware AI agent that engages the user in conversation during their daily routines and practices


Based on the feedback, we decided to move forward with combining elements from Concept 1 and Concept 3. We revised our concept to be an AI Home Agent that used different medias (voice and visual) to provide the user with information within a home setting. We also incorporated the idea of a social network that would connect multiple users of our device to each other. This would provide a platform for citizen-driven advocacy within neighborhoods. 


Concept Features + Speed Dating

Based on our concept, we developed various concept features which we speed dated. We used the following storyboards as stimuli to engage in conversations with users. We focused on inquiring about:

  1. The level and frequency of interactivity

  2. The degree of integration within the home

  3. The balance between voice vs visual vs touch interactions

  4. The comfort level of communication with strangers or neighbors through a social network

  5. Gamification as a motivating tool

Storyboards for Speed Dating_Page_1.jpg

Key Insights: Evaluative

  1. People were less inclined to interact for long periods of time with information that is scattered throughout the home.

  2. There must be a balance between the device being passive and active in its degree of interactions.

  3. People desired a central location that they could interact with information at their own pace.

  4. Goals were a positive way to motivate users.

  5. People liked the idea of suggestions from a device to make their home more efficient.

  6. People liked the idea of suggestions about environmental organization, but only if it was personalized to their interests.

  7. People were concerned about privacy and were not comfortable with the idea of seamless communication with their neighbors through an IoT device.

Based on the speed dating findings, we revised and finalized our concept and design principles. We found that we were trying to make our device do too much and the connection between the home monitoring, social network, and environmental organization suggestion platform was not aligning. Because of this misalignment, we decided to focus on the integration of the device within the home and the incorporation of the environmental organizations.

Final Design Principles

  1. Our design should focus on small scale, daily life activities.

  2. Our design should make the impact visible.

  3. Our design should use awareness as an empowering catalyst for behavior change.

  4. Our design should connection personal values to a larger environmental context.

  5. Our design should be flexible and allow for personalization based on the user's daily routine.

  6. Our design should motivate the user but through personalized means.

Final Concept

An in-home IoT device that monitors 4 utilities (energy, gas, water, and air quality) and relates those utilities to local sources to promote sustainable behavior change. In addition, it matches users to environmental organizations based on the user engagement patterns.  


04 Finalizing the Details

Once we identified the features we were going to prototype, we began finalizing those details. These included ambient visualizations, projected visualizations, and the form of the device.


Mental Models Sketching

For the ambient visualizations, we wanted to utilize metaphors to represent the different utilities. To gain an understanding of mental models, we had users draw visual representations of the 4 utilities we were addressing. 

In addition, we had users map out a typical day and identify when they would interact with our device. We found that people wanted the ability to access information on the go, which resulted in developing a mobile app for our system.   


Developing Metaphors

We used the results of the mental models sketching to inform the metaphors we used to visualize the four utilities. We identified the different metaphors and accompanying visual variables. 

  1. Gas: Black particle cloud (metaphor); Height + Density (visual variables)

  2. Water: Water vessel (metaphor); Height + Volume (visual variables)

  3. Energy: Lightning (metaphor); Intensity + Color (visual variables)

  4. Gas: Stove flame (metaphor); Height + Color (visual variables)


Form Integration

We considered how the form of the product would work with the ambient and projected visualizations, as well as design details, including material choice and small design features, such as the incorporation of an on/off mic button. We also considered the scale of the product by looking at where the device might be placed within a home. 


05 Incorporating Learning

In addition to the design of the device, we wanted to map out how the learning component would work within the Orion system. We developed personas of our target user and the AI agent to gain an understanding of the relationship between them. 

I took the lead on structuring the learning component of the Orion system and developing the frameworks based on research I performed about different learning theories.


We developed a learning framework that focused on identifying the goalposts that Orion would work to move the user through. The framework is designed in a way that it can be personalized for each user. Each of the 5 goalposts are separate nodules that build on one another, however based on the desired goals of the user Orion will adjust the speed and movement through these goalposts.

For example, one user's goals may focus on creating an energy efficient home, so Orion may only move this user through nodule 1 and 2. If Orion finds that the user starts to express interest in topics related to the other nodules, based on the interaction patterns between the user and Orion, Orion will test the user and begin to move them to the next nodule. 

Learning goalposts.png

Because motivating was an important element to the success of Orion, we developed a motivation framework that would use multiple tactics to engage and motivate the user as time goes by. This framework is based on spontaneity. Orion uses a set of motivators (sound, light, movement) to engage the user when they are exhibiting low engagement based on a motivation profile Orion develops about them. This profile is periodically reevaluated over time to continuously make adjustments to the motivation tactics. In addition, Orion uses topics of interest to the user, for example money or health, to motivate them to exhibit changes in their behavior. 


05 Future Considerations

This project is an exercise in experiencing deep investigation and research and how it informs the design of a solution. With my interest in design research, I found the experience of structuring research plans and facilitating research methods very fulfilling and a great learning experience. For next steps, because we are dealing with behavior change, we need to user test over longer periods of time. In addition, we would like to explore the volunteer organization portal and how it would function and connect with Orion more in-depth.