Online Traffic Signal Control through Sample-Based Constrained Optimization

Srishti Dhamija, Alolika Gon, Pradeep Varakantham, William Yeoh

PosterID: 65 PDF BibTeX

Traffic congestion not only reduces the productivity of individuals (by increasing the time spent in traffic) but also increases pollution. Controlling traffic signals effectively can significantly reduce traffic congestion and hence has received much attention from researchers. To better handle changing traffic patterns (and avoid situations where vehicles travelling in one direction wait at a traffic signal even when there are no vehicles travelling in other directions), recent work has focussed on online traffic signal control. Typically, the objective in traffic signal control is to minimize expected delay over all vehicles given the uncertainty associated with the vehicle turn movements at intersections. In order to ensure responsiveness in decision making, the leading approach for online traffic signal control computes a schedule that minimizes the delay for the expected scenario of vehicle movement instead of minimizing expected delay over the feasible vehicle movement scenarios. Such an approach degrades schedule quality with respect to expected delay as the probability of vehicles taking turns at intersections increases. We introduce TUSERACT (TUrn-SamplE-based Real-time trAffic signal ConTrol), an approach that minimizes expected delay over samples of turn movement uncertainty of vehicles. Specifically, our key contributions are: (a) By exploiting the insight that vehicle turn movements do not change with traffic signal control schedule, we provide a scalable constraint program formulation to compute a schedule that minimizes expected delay across multiple vehicle movement samples for a traffic signal; (b) a novel mechanism to coordinate multiple traffic signals through vehicle turn movement samples; and (c) a comprehensive experimental evaluation to demonstrate the utility of TUSERACT over the leading approach for traffic signal control, SURTRAC. Our approach provides substantially lower (up to 60%) mean expected delay relative to SURTRAC with very few movement samples while providing real-time decision making.
Session Am2: Scheduling & Traffic Control
Canb 10/29/2020, 03:00 – 04:00
10/31/2020, 10:00 – 11:00
Paris 10/28/2020, 17:00 – 18:00
10/31/2020, 00:00 – 01:00
NYC 10/28/2020, 12:00 – 13:00
10/30/2020, 19:00 – 20:00
LA 10/28/2020, 09:00 – 10:00
10/30/2020, 16:00 – 17:00