Research on Driver Behavior Index Wins Big at Data Science Day

Patrick Alrassy
Patrick Alrassy Photo by: Timothy Lee

Patrick Alrassy, a doctoral student affiliated with the Smart Cities Center at the Data Science Institute (DSI), won the Best Demo Award for his research presentation on developing a “Driver Behavior Index” during the 2018 Data Science Day. 

Alrassy is part of a team that’s creating a Driver Behavior Index by gathering data from GPS devices mounted on a fleet of 27,000 city-owned vehicles. Their work, funded by New York City Department of Transportation (DOT), offers a snapshot of drivers behavior by accumulating data on speed corridors, traffic congestion, acceleration and hard-braking spots on a network of roads across the city. The GPS devices contain mobile vehicle embedded sensing devices that collect the data.

The research team includes Alrassy’s doctoral adviser, Andrew W. Smyth, professor of Civil Engineering and Engineering Mechanics and chair of the Smart Cities Center at DSI, and Jinwoo Jang, a former doctoral student at Columbia Engineering who is now an assistant professor at Florida Atlantic University.

“Once completed, the Driver Behavior Index will assist city agencies in planning and implementing NYC Vision Zero, an initiative to end traffic deaths and injuries on city streets,” says Alrassy. “Compared to immobile radars, data from GPS devices mounted to a fleet of vehicles provide more accurate analysis of road-network performance.”

The data from the GPS devices, though, is often “noisy” or corrupted by “dense urban environments,” Alrassy adds. The team therefore developed a map-matching algorithm that can infer the trajectory of vehicles. The algorithm is capable of matching each raw data point from the GPS to the road where the vehicle has most likely traveled. Also, after they process the raw data, each road segment has a set of data points. From this data, they can learn the statistics of each road segment, such as inferring the speed profile on a given road segment across the day (the a.m. peak versus the p.m. peak versus off-peak hours).

Their results are then fed and visualized in ArcGIS, a geospatial platform, which shows the variation of each statistical parameter in the Driver Behavior Index across all New York City corridors.

Alrassy demonstrated the team’s research during Data Science Day by using three large computer monitors. There were a total of 19 demonstrations and 42 posters by undergraduates, graduate students and postdocs. The 700 participants at the conference affiliated with Columbia were invited to judge and score the presentations and voted his the Best Demo. He started his demonstration with a PowerPoint presentation, followed by a sample run of his map-matching algorithm and a visualization of the map-matched data on Google Maps. He also illustrated the speed corridors on the ArcGIS platform. 

“It’s a great honor to work on a project that uses the most advanced techniques to help make cities safer,” said Alrassy. “Our research merges civil engineering with data science in a way that will help cities implement smarter transportation policies and become safer.”

 

--By Robert Florida


550 W. 120th St., Northwest Corner 1401, New York, NY 10027    212-854-5660
©2018 Columbia University