Project Overview
Spring 2022
Duration:
8 weeks
Field:
IoT, Intelligent Environment, Interaction Design, Experience Design
Tools:
After Effect, Arduino, Touch Designer, Fushion 360.
For this project, I was tasked to create a design for the TCS building on campus at CMU. The building is equipped with IoT devices that have 9 physical sensors. Data are collected, stored, used and shared. However, the data that they collect remain intangible for the occupants in these spaces.
So, How might we improve data awareness & transparency in university building?
My design proposal is called DataHive, which is a fantastical data interpreter that harvest data from IoT sensors to generate stories and create space for people’s imagination to grow. It invites people to co-create the legacy of the space and play with collected data in the form of puzzle pieces.
So, How might we improve data awareness & transparency in university building?
My design proposal is called DataHive, which is a fantastical data interpreter that harvest data from IoT sensors to generate stories and create space for people’s imagination to grow. It invites people to co-create the legacy of the space and play with collected data in the form of puzzle pieces.
Problem Space
Considering that more and more universities start to install IoT sensors in learning environments that gather huge amount of data...
![]()
How might we improve data awareness & transparency in university buildings?
Location
I picked the RI Master’s Commons on the fourth floor in TCS Building as my location to work with. There is a matrix of sensors embedded in the ceiling, but the only thing it controls is turning the light on and off, which is seems to be an underemployed usage considering the variety of data these sensors can gather.
About The Sensors
Mites sensor integrates a myriad of sensing capabilities, minus a camera. There are nine physical sensors capturing twelve distinct sensor dimensions.
![]()
The middleware will process data gathered by Mite sensors and use ML to recognize “Events” such as “light off”, “people walking”...
Observation
After observing and interviewing people who work in this space, I formed an initial understanding who what they need and what are the unique features of this space.
- Longterm occupancy: This space will be occupied for a long period of time throughout the day as a workspace.
- Social Interaction: People who come to this space know each other but feel discouraged to socialize due to the layout of the room and lack of oppurtunities.
- Exclusivity: This space belongs to RI masters so there are less variables regarding the source of data.
Values
Data Transparency
Reveal the black box of IoT system in an comprehensible way.
Appropriation
Enable users to adapt this IoT system to reflect themselves.
Foresight
Create meaningful legacy of the space for people to make long term connections.
User Journey Map
A good interaction design should feel like a continuing conversation that intrigue and inspire people to look at the same thing differently. Data Hive interperates data like storyteller following a set of rules that makes it engaging and relatble.
Users will gradually unfold the potential of these stories through exploring correlations. I believe being intrigued to read data is the first step toward better data transparency.
Space Diagram
The spatial layout of each component of the experience.
System Diagram
This is how the technology works!
For example:
Raw Data
(from sensor)
💡️light intensity: 0
+
🚶♀️Motion: 1
(from sensor)
💡️light intensity: 0
+
🚶♀️Motion: 1
Event
(from ML)
“Room turned Dark”
+
“People Walked”
(from ML)
“Room turned Dark”
+
“People Walked”
Text
(from AI writer)
“Dark as coal mine”
+
“Two ___ walked into ___”
=
One long winter night, two ___ walked into ___ as dark as coal mine...
(from AI writer)
“Dark as coal mine”
+
“Two ___ walked into ___”
=
One long winter night, two ___ walked into ___ as dark as coal mine...
Jigsaw Piece Breakdown
Each jigsaw piece carries a story about the space and the people. On the front side, you will see AI-generated stories based on the events recognized from the Mites middleware such as "people walked" and "Room turned dark." These events will be bolded to show what the system can infer from the raw data. Clearly showing what the system knows or doesn't know will improve its transparency, while leaving the blanks invites occupants to participate.
