Modeling municipal vulnerabilities to bolster climate resilience.
Duration
5 Months (Aug - Dec 2018)
Team
Jeong Min Seo, Tyler Stern, Allana Wooley
Tools
Figma, Principle, Final Cut Pro
My Role
Prototypes (Lo-, Mid, & Hi-Fi), think aloud testing, animation, & videography
$4.3 billion
The average amount of damage caused by floods per event.
These disasters will only worsen with time as climate change drives sea level rise, increasing the threat of powerful storm surges and heavy rainfall.
Further, experts' current tools are insufficient to meet the demands that the next generation of floods will bring.
An intelligent mapping software that uses real-time data to analyze existing infrastructure and predict flooding likelihood.
Using data collected by KAARTA's LIDAR devices mounted onto city vehicles, TRITON creates a common platform of consistently updated information for key stakeholders (such as city governments or disaster relief organizations) to collaborate and coordinate efforts.
City vehicles collect environmental data
Computer Vision flags damaged infrastructure
City official reviews collaborative map
Desktop Portal
A cross agency collaborative platform that displays flagged structures and data layers stored in the cloud.
Notes from Different Data Layers
Both officially recognized government organizations and general public have access to TRITON, designated by differently coded notes.
Storm Mode
Predictive Modeling of Flood Areas based on precipitation levels and incoming forecasts.
Localized Alerts
Push alerts for subscribed users in affected areas encourages transparency and public participation.
The research methods employed during our generative phase:
Literature Review
Reviewed academic papers and news articles on computer vision
Expert Interviews
Interviewed civic engineers adept with mapping technology
Analogous Domains
Compared similar design patterns in other problem spaces
To begin, our team needed to understand the strengths, shortcomings, and applications of Computer Vision. We split up to research the topic and ultimately synthesized our findings into three areas of potential application: Forensics, Disaster Response, and Healthcare.
We noticed an intersection between Forensics and Disaster Response, particularly in civic environments.
Given the importance of climate resilience, would there be a problem space for utilizing CV to assess current infrastructure for disaster risks, such as flooding potential?
We chose to investigate this question further.
We dug deeper into the topic of modeling civic infrastructure by interviewing two civil engineers. We wanted to learn about their day-to-day operations, mapping tools, and general pain points.
Civil Engineer
Sherwood Engineering
It is "massively time consuming" to create flood models of a site and properly map data.
Civil Engineer
U.S. Army Corps of Engineers
Water and flood controls systems are inspected only "once every three years."
"Even year-old maps are out of date due to the speed of climate change."
From these interviews it was clear that a big pain point in disaster management was updating maps and models on affected areas.
Currently, many civil engineers and urban designers use tools like ArcGIS. However, the Civil Engineer from Sherwood Engineers describes these tools as "inaccessible" to all but expert users.
Given KAARTA's proficiencies, developing accessible tools to model floods seemed like an excellent design opportunity.
Alongside our interviews, we looked at some unconventional inspiration in "analogous domains." The idea of scanning environments for surface features to glean deeper insights was a similar feature to Batman's "detective mode" in the Arkham series of video games.
Not unlike Computer Vision technology, Batman could active a scanner that would identify an item or surface of interest, scan it for deeper information, and then perform an array of magically inspired analysis like recreate bullet casings, track footprints, or in the coolest way, reconstruct actual events that had transpired.
Most of this is of course science fiction, but the video game inspired us to look at 4D models of environments grounded in scanned data.
MIT's Media Lab features a project called CityScope that allows users to create interactive models of cities with legos. As users build and break apart certain bricks, a camera registers the particular type of brick and a projector accordingly depicts the part of the urban design that the user is trying to emulate. These interactive models would ultimately inspire our first prototypes.
Computer Vision + Civics
Opportunity exists for Computer Vision in civic sectors.
Technical Shortcomings
Modeling flood zones is costly, time-consuming, and always one step behind nature.
Urban Visualization
Visualizing useful urban and civic data is not limited to 2D forms.
Inaccessibility
It is hard for the public to participate with esoteric mapping tools and data.
When it came time to start prototyping, our professor handed us metal wire and told us to get ideating. Given that it was likely our final deliverable would be a digital product, this was a unique challenge.
Since we couldn't make anything related to screens with the wire, I proposed the idea of an Augmented Reality system where users could interact with a fully scalable 4D depiction of a city block.
Similar to Batman's "Detective Mode" and MIT Media Lab's CityScope, we envisioned this interactive model featuring colored pins that noted infrastructure damage and flooding hazards.
Our body-storming/experience prototyping exercise revealed potential for interaction with otherwise cumbersome data. It also provided us with key insights for further prototypes that would be required to enable this sort of interactive system. We drafted a project roadmap to capture our ideas.
As cool as our 4D interactive map was, we couldn't find a way to create a prototype of high enough fidelity to test with users given the time restraints of our project.
In addition, we noticed this sort of collaborative system would require a cloud service to store data in a centralized place so that different users could access the same data layers using different platforms. Luckily, KAARTA already offers this kind of service.
Therefore, what made most sense for us to continue prototyping was a 2D digital layer similar to ArcGIS or other mapping software. We could easily prototype screens for this idea and test them through Heuristic Evaluations and Think-Alouds.
Our initial prototypes were created through collaborative sketching. We constructed a 2D digital layer as noted in our roadmap.
We mainly focused on visualizing notes added to the map by different stakeholders as well as Storm Mode: a predictive view that would display flooding potential and estimated damage in areas based on different levels of precipitation.
Because we didn't have access to our engineers from our interviews, we couldn't test any lo-fi prototypes with them. Therefore, we moved ahead to creating Mid-Fi prototypes to test the heuristics of our platform.
Evaluative Research Goals:
Both novice and experienced users found the amount of information on the background map layer overwhelming.
Experienced users needed notifications to integrate with existing work flows (e.g. email).
Because we didn't have access to our engineers from our interviews, we couldn't test any lo-fi prototypes with them. Therefore, we moved ahead to creating Mid-Fi prototypes to test the heuristics of our platform.
Our final iteration was presented to our class along with our concept video. Unfortunately, we were not able to present our project to KAARTA due to communication failures between our professor and the organization.
It was a bummer we couldn't present our final iteration to KAARTA. Our team was interested in getting feedback, particularly regarding our idea to partner with city governments. In addition, we were unable to get back in touch with our Civil Engineers that we spoke to during our Discovery phase. We would have been interested to hear their thoughts on our prototypes. Understanding if our desktop portal had any usefulness in the current ecosystem of the municipal government's technical tools would have been our research goal moving forward.