InTrans / Jan 03, 2019

REACTOR Lab students finalists in TRB competition

REACTOR Lab logoA team made up of REACTOR Lab graduate students was one of five finalists in a transportation forecasting competition led by the Transportation Research Board (TRB).

The team consisted of Archana Venkatachalapathy, Pranamesh Chakraborty, Subhadipto Poddar, and Tongge Huang, who work at InTrans’ REACTOR Lab under the guidance of Dr. Anuj Sharma.

“We’re really excited and proud of these students and the recognition they received based on the incredible opportunity provided them through working on real-world problem solving for the Iowa DOT Operations Bureau,” said Neal Hawkins, co-director of the REACTOR Lab and InTrans associate director.

The international competition challenged the teams to use a massive data set from DiDi Chuxing Technology Co., a Chinese ride-sharing company, to predict traffic data as close as possible to real-world data. The teams had to develop traffic forecasting models to predict the average speed at five-minute intervals on a road section along Chang’an North Road in Xi’an, China.

The Reactor Lab students finished fourth out of 31 teams competing from across the US and internationally. The top five teams will go on to compete in the next round. They will present their work during a TRB Annual Meeting workshop on big data.

Shuo Wang, an Iowa State graduate and former REACTOR Lab student, developed the hybrid long short-term memory (LSTM) prediction model that the team used. The teams’ predictions were evaluated by root mean squared error to determine the most accurate models.

The competition was organized by the TRB committee on Artificial Intelligence and Advanced Computing Applications (ABJ70) and Institute of Electrical and Electronics Engineers’ (IEEE’s) Intelligent Transportation Systems Society (ITSS) Technical Activities Sub-Committee “Smart Cities and Smart Mobility” and sponsored by DiDi Chuxing, a Chinese ride-sharing and technology firm.