So a while ago, we started working with a small group of research partners in the EU and the US to see if we could tackle some of these challenges together, using aggregate, anonymised data about historical traffic patterns to help improve urban mobility for everyone.
Our initial exploration has lead to a series of pilot projects with our partners to find ways to minimize traffic congestion, speed up journeys, improve safety, and reduce the amount of money spent on infrastructure.
In Stockholm, a city with many bridges and tunnels, we’re working with KTH Royal Institute of Technology to reduce the number of tunnel closures on the Södra Länken, the second longest urban motorway tunnel in Europe, to improve travel times for citizens.
In the Netherlands, we’re working with the Netherlands Organization for Applied Scientific Research (TNO) to see whether it’s possible to reduce their reliance on expensive physical road sensors for information about traffic flows. The aim is to reduce infrastructure costs without compromising on traffic safety. We’re also working with the Amsterdam Institute for Advanced Metropolitan Solutions on related questions.
We’re also working with major research institutions and transportation planning groups in Denmark (the Technical University of Denmark) and in the US (the Rudin Center for Transportation at New York University Wagner School of Public Service, San Francisco County Transportation Authority, and the University of Chicago Harris School of Public Policy).
We only share aggregate, anonymized snapshots of historical traffic statistics with these institutions, including average traffic speed, relative traffic volumes, and traffic trajectory patterns. These statistics are derived from aggregate Location History data that our users have proactively chosen to share with us (and which they can switch off again at any time via My Account). This is the same data we use in Google Now to notify users of disruptions to their commute due to traffic, and tell them about the best time to visit their favourite museum in Google Maps.
To ensure that no individual user’s journey can be identified, we only share representative models of aggregate data employing a technique called differential privacy, which intentionally adds “noise” to the data in a way that maintains both users' privacy and the data's accuracy. The technique has also been successfully tried and tested in Chrome.
It's still early days, but preliminary results have been positive. In the Netherlands, TNO ran tests on a 10km stretch of highway that regularly faces traffic jams, using our anonymized traffic statistics instead of physical road sensors. They found that they could still accurately detect traffic jams at the right moment and at the correct location on the road without the sensors, potentially saving 50K Euro per year if the redundant sensors were removed. Other pilots are starting to show similarly positive results.
We’re excited by the promise that these initial projects have shown in meeting the challenges of urban mobility, and today, we’re pleased to announce that we’re expanding our pilot programme. We’re seeking to build partnerships with cities and research institutions to evaluate ideas and run experiments, ultimately improving urban mobility for everyone. If you’re working on a project addressing congestion, pollution, safety and similar mobility challenges, and are interested in working with us, please get in touch.