2020–2022 · STE–NUS Joint Laboratory
Urban traffic flow smoothing
This industry collaboration developed dynamic traffic-information strategies for road incidents.
Research question
How should traffic information be updated after an incident so that drivers’ route and lane-changing responses help the network recover?
Approach
I worked on a deep reinforcement learning system that integrates historical and real-time data with a calibrated 29-kilometre PTV Vissim model of the Serangoon road network in Singapore.
Contribution
The project connected online learning with traffic microsimulation and produced a software package incorporated into ST Engineering’s urban mobility solutions.