Designing streets for safety
Artificial intelligence can prove old theories.
There are long-held, accepted theories about what makes a neighbourhood healthy. Until recently, though, they’ve been just that – theories. The use of data and Artificial Intelligence (AI) now makes it possible to test them. In a new study by Cesar Hidalgo, an MIT Media Lab associate professor, and coauthors Marco De Nadai and Bruno Lepri, experiment with two well established theories about urban design.
One is Jane Jacobs’ natural surveillance theory. Put simply, it’s based on the idea that isolation makes crime both easier and more likely. Jacobs says elements like street lights, windows, and open spaces help residents keep their neighbourhoods safe.
And Oscar Newman’s defensible space theory is the idea that architectural details like arches or steps create semi-private spaces that residents are more apt to watch and defend.
Hidalgo, De Nadai, and Lepri used a trained neural network to analyze tens of thousands of StreetView images in Milan and Rome. They then compared those with cell-phone data. What they discovered proved Jacobs’ and Newman’s theories. There is, in fact, a definite relationship between safe-looking streets and activity, and that “people over 50 and women were more likely to be active” in neighbourhoods that are perceived as safe.
“There is a definite relationship between safe-looking streets and activity, and people over 50 and women are more likely to be active in neighbourhoods that are perceived as safe.”
It’s important to note that a relationship between perceived safety and aesthetics doesn’t equal a relationship between aesthetics and crime. But understanding the elements with positive associations –– green spaces, open areas, glazing, sidewalks –– will certainly influence urban planning and design.