We held a day-long fall hackathon. Here’s what we built
Recently, our engineering team has thought a lot about ways to improve quality of life in global cities like Toronto. But we know that much of the growth in cities over the next few decades will come from developing regions — with India, China, and Nigeria accounting for more than a third of urban growth by 2050, according to the U.N. So we wanted to explore ideas to apply technology to challenges in emerging areas, even in a limited, preliminary way.
To get started, we set up a day-long hackathon with some ground rules. Each team started by researching an established challenge in emerging cities, and considering the types of tools that might help address some aspect of it. The teams then pushed to create a functional prototype within a 24-hour period. The goal wasn’t to produce a polished demo, but rather something tangible to show us what might be feasible in the future.
Let’s be very clear: we don’t have answers to the enormous challenges facing developing cities, nor do we think a one-day sprint can make a dent. But what hackathons do well is create fresh opportunities to consider tough problems with new tools — and help pose better questions to the people who study these problems every day. Here’s what we came up with.
Team: Michelle Ha Tucker, Tom Kennedy, Dan Riegel, Kristine Sarnlertsophon
More than 90 percent of global traffic deaths occur in low- or middle-income countries without advanced road infrastructure. While many factors are involved in these deaths, from street designs to car-based planning, our team focused on traffic signals. Specifically, we asked ourselves: Could we build low-cost traffic lights that can improve safety but that don’t require grid power, expensive technology, and extensive physical infrastructure?
The resulting concept was a low-infrastructure system called Sol Signal. Powered by a solar panel, each Sol Signal uses low-power PyCom controllers to govern the light sequence and a low-power, low-bandwidth network (also known as LoRAWAN) to communicate with other signals in the network.
In contrast to existing traffic signal systems — which depend on expensive, disruptive underground wiring for light coordination and power — this low-infrastructure approach means each set of four signals can be created for less than $1,000, making it more feasible for emerging areas to deploy at a scale that can save lives.
Visualizing Climate Change in Guangzhou
Team: Jack Amadeo, Matt Breuer, Katherine Guo, Samara Trilling
Emerging areas are especially vulnerable to the potential impact of sea level rise. Describing the threat facing emerging Chinese coastal cities, for instance, the New York Times wrote in 2017 that the port city of Guangzhou “now has more to lose from climate change than any other city on the planet” economically speaking, not to mention the human cost.
Models can help cities prepare for climate threats, but creating reliable models can be a challenge. To explore the potential for models in Guangzhou, this team used an internal tool that visualizes cities alongside a number of data sources — namely, elevation data via a Google Maps API, and buildings and water data via OpenStreetMap.
The gif shown above — based on sea level rise up to 7 meters — begins to convey areas that might be at risk. A big caveat: The tool’s accuracy is limited by the available data; for example, we didn’t have strong metrics on building height or water absorption. But if further developed, such a tool could help allocate resources for protection, shape development patterns, and communicate risks to the public.
Health Planning for Refugee Camps
Team: Marie Buckingham, Betty Chen, Brian Ho, Craig Nevill-Manning, Ananta Pandey, Violet Whitney
Refugee camps are created rapidly in response to conflicts, but they often grow beyond their planned capacity or face resource limitations, leading to challenges around care or services. For example, at Za’atari, the world’s largest Syrian refugee camp, sexual health problems are the leading cause of death.
This team hypothesized that it might help improve access to care, in a climate of limited funding, by identifying places to distribute low-cost resources such as educational materials, birth control, and menstrual health supplies. Using imagery of the camp found online and an internal planning tool called generative design (more on GenDes in a future post), this team first predicted busy thoroughfares in a 0.2-square-mile area of the camp. Then it mapped public places where women currently gather, such as latrines and schools, but that are currently unrelated to sexual health.
Locating low-cost resources in high-traffic locations could help women access support in their day-to-day activities, before they have a more serious problem that requires visiting a clinic. The challenges of providing care in refugee camps go way beyond planning into much larger issues of funding and safety, but knowing where to direct limited resources could still improve access to resources that support women’s health.
Transferring Files Without Data
Team: David Huang, Okalo Ikhena, Douwe Osinga, Stefanie Pitaro, Rachel Steinberg
Access to the internet remains a challenge in many cities around the world, particularly in emerging areas. And even in connected cities, mobile data plans can be expensive.
This team took aim at the digital divide by hacking together a system capable of transferring image, video, and audio files offline. The system enables file-sharing across lower-cost smartphones that use the Android operating system, often found in emerging areas. This file-sharing could repurpose the phone’s Wi-Fi module to rapidly share data with other phones — within a range of roughly 100 feet — without connecting to the internet.
While our system didn’t actually share files, it did discover other devices in its range, which helps prove its technical feasibility. To take this system to the next level, this team would need to develop real use cases and consider many complicating factors (e.g., what if lots of files were being shared at once?), and advance the tool accordingly.
Team: Nick Jonas, Jeremy Neiman, Dan Vanderkam, Natalie Weires
The design of street grids has a major impact on transportation systems, job access, and urban development more broadly. Ideally, street networks in emerging areas could adapt some existing best practices that best fit local needs.
In that spirit of capturing insights from established street grids, this team pursued a Street Sketch tool that could learn to draw different styles of street grids, and potentially even apply those styles to another geography. So, using a tool called sketch-rnn, we trained Street Sketch using 10,000 examples of street grids.
Admittedly, as the gif above suggests, this hack didn’t work out as planned. We’re not quite sure why: it could have been insufficient time or training data, the wrong data format, bad training parameters, or some combination of factors. But a failed hack is valuable in its own right—in this case showing us just how much more preliminary work is needed to pursue something like Street Sketch in the future.
November 1, 2019