Traffic at the intersection of sensor data and social media

Dublin has one of the most advanced intelligent traffic systems in Europe. The city is full of sensors at its intersections (induction loops counting cars, remotely controlled traffic lights, and traffic cameras) and on its bus fleet equipped with on-board satellite positioning units, all streaming data in real-time. Now, it is turning to crowdsourcing. The city is plugging social media data into its INSIGHT System, powered by research done at IBM Research-Ireland. Two technologies use data from the commuting public to help urban traffic controllers get a better view of city roads to more-quickly respond to incidents, and issue alerts, thereby improving the accuracy of the Insight system and the availability of traffic alerts for Dubliners.

Jakub Marecek
As IBM Research-Ireland’s Jakub Marecek and his numerous co-authors from the INSIGHT project explain in their paper Heterogeneous Stream Processing and Crowdsourcing for Urban Traffic Management, at the International Conference on Extending Database Technology, crowdsourcing is as a tool for resolving ambiguity in intelligent transport systems. 

"It queries volunteers close to the sensors that disagree (with reports) and estimates what has actually happened, considering the reliability of participants' responses so far.

“The benefits of this approach are two-fold. First, more-accurate information is given directly to end users. Second, the complex event processing component of our system makes use of the crowdsourced information to minimise the use of unreliable sources. The traffic modelling component may also use the crowdsourced information to resolve data sparsity.”

Alerting the crowd

One part of the system is the CrowdAlert app, which ask its users for voluntary confirmation of traffic incidents and additional information. The free Android app is available on Google Play. When an unusual event, such as a major slow-down, is observed in the induction loops and bus position traces, the users are alerted and, if nearby, are asked for their impression of the causes. In return, the users see a variety of alerts and traffic data on a map-based interface. The Dublin City Council, which manages the system, may also ask users further questions and offer award, such as parking credits for the answers.

“The CrowdAlert app is city agnostic so it can be used in any city. However, the server handling the data for the application would need to be adapted depending on the types of sensors used in that city,” Jakub said. Warsaw, Poland is now considering a CrowdAlert deployment.

Crowdsourced tweets on the radio

A prize-winning paper in Association for Computing Machinery Recommender Systems Conference, Crowd Sourcing, with a Few Answers: Recommending Commuters for Traffic Updates,  by IBM Research-Ireland's Elizabeth Daly and Michele Berlingerio, and their colleague Francois Schnitzler, Senior Researcher at Technion – Israel Institute of Technology, details an algorithm that predicts the best Twitter users to follow or query about traffic in Dublin.

It crowdsources traffic updates based on user engagement level and reputation, which combines the time, and mobility profile of the user; the time of day and locations the user typically communicates about; and the number of queries previously directed at the user. The team evaluated this proposed strategy using a collection of user contributed tweets and live traffic updates from Dublin City Council’s Dublin City FM radio station @LiveDrive account.

The station's DJ broadcasts LiveDrive tweet updates during morning and evening peak traffic times, after a producer monitors incoming alerts, and a camera controller, who has access to a number of the traffic cameras in the city, clarifies reported issues. Listeners are incentivised to provide real-time updates, via their own Twitter accounts to the station, with the possibility of receiving free parking vouchers, or a song request – not to mention, the general understanding that if they contribute useful information that benefits others, they too may benefit from the contributions of others. 

Using the LiveDrive tweets as the gold standard for other tweets, the researchers were able to build and analyse the reputation of other users who tweeted about traffic. They collected almost 40,000 tweets originating from and directed to the LiveDrive account by almost 2,200 accounts from October 2012 until February 2014.

There has been an incident on Nassau St at the junction of Dawson St. Nassau St is blocked and the left turn from Dawson St is also blocked.
— Live Drive (@LiveDrive) February 14, 2014

Elizabeth Daly
“We were amazed by the participatory model set up by the radio, which involves user by both broadcasting tweets about traffic, and by receiving alerts from them,” said Elizabeth.

“We then wondered: what if LiveDrive had a way to directly contact people using Twitter and ask them about updates on traffic and accidents?”

The researchers were able to accurately predict which Twitter user was most appropriate to provide information, depending on the time of day and location of interest.

“The work is in an experimental stage but speaking to the LiveDrive team they expressed interest in understanding the reputation of the users contributing updates saying some users ‘are very loyal contributors and we really know we can trust their content. Having a way to see an indication of history of the person contributing the update would really make that easier.” wrote Elizabeth in her paper Westland Row why so slow?: fusing social media and linked data sources for understanding real-time traffic conditions.

IBM’s partnership with the Dublin City Council shows that Twitter data and crowdsourcing can successfully add commuters’ voices to the city’s transportation systems. Traffic managers will know about incidents sooner so they can provide commuters with realistic alternative routes and more accurate travel times.

Labels: , , ,