Artificial Intelligence for Healthy Transportation Planning
How our smartphones could be helping planners build safer and healthier cities
By Avipsa Roy, PhD Geography, Arizona State University
Smartphones, wearable devices, and fitness trackers are driving a data boom that can help us understand how people move around in cities, how we can build our cities to be safer, fairer and more active.
You and I now act as sensors thanks to our smartphones: billions of location data points are recorded all over the world every day. The challenge is no longer about getting the data, but about finding useful information from the raw data. To do this, researchers are using artificial intelligence algorithms to enable planners to harness the power of big data for urban planning decisions.
The idea behind AI is that it allows us to automate processes with very little human intervention. For example, AI algorithms can find optimal routes for different transportation modes and predict where people using different modes will go. The algorithms do this by learning from existing data, such as historical mobility patterns to improve future travel.
AI can go beyond optimizing simplistic mathematical criteria, like finding the shortest route between two places. It can recommend personalized routes that fit the needs of individual cyclists based on their data. These algorithms can think as a cyclist with specific preferences, like wanting to avoid hills, use protected bike lanes, or pass by a coffee shop.
Using AI algorithms for personalization of user choices is already common in online advertisements and retail. However, not many AI algorithms improve route recommendations for bicyclists, pedestrians, and motor vehicle users.
Thanks to our smartphones and apps like Strava, researchers can access detailed movement data. Strava records routes of tens of thousands of bicyclists in a city each day. These can be mapped to see where people ride the most. Although the data is biased towards recreational bicyclists, using suitable correction factors can make it more usable for AI algorithms. And citizen science tools like Bikemaps.org can track safety of bicyclists in a neighbourhood. We can combine these sources of information to add context to transportation mode choices for residents and improve AI algorithms.
I’ve been working to develop models to make better predictions for the safest, most comfortable and popular routes for pedestrians and bicyclists in a neighbourhood. If somebody wants to find the safest way to bike across a canal as well as the most comfortable cycling experience — it may soon be possible with just a tap on the screen of your smartphone. I have also been working with INTERACT researchers to determine someone’s mode of transportation based on their smartphone data.
These increased sensing capabilities and the power of connected transportation infrastructure raise concerns about privacy and data safety, which can’t be ignored by policy makers and technology companies in the years to come.
Avipsa Roy is a PhD candidate at Arizona State University who focuses on harnessing the power of AI to improve safety, access, and comfort for cyclists and pedestrians.