ITS Berkeley

Traffic congestion is forecast for neighborhoods within a region using a deep learning model. The model is based on Long Short-Term Memory (LSTM) neural network architecture. It forecasts a congestion score, defined as the ratio of the vehicle accumulation inside a neighborhood to its trip completion rate. Inputs include congestion scores measured at earlier times in neighborhoods within a region, and three other real-time measures of regional traffic. The ideas are tested using Newell’s simplified theory of kinematic waves. Simplified street networks are featured first. Initial tests...

Two new synchronization strategies are developed for signalized grids of two-directional streets. Both strategies are found to reduce congestion significantly more than do other approaches. One of the strategies is static and the other adaptive. Both use a common timing pattern for all signals on the grid but use a different offset for each. The static strategy serves the morning rush by providing perfect forward progression on all streets in the directions that point toward a reference intersection, one that is located near the center of gravity of all workplaces. For the evening rush,...

The paper develops a novel strategy for delivering feeder service in support of trunk-line transit. The strategy is well suited to developing countries, where costs of emergent communication technologies often preclude their use. The strategy, termed Jitney-lite, is a form of collective transportation that provides a degree of flexibility. Patrons who board an outbound Jitney-lite vehicle at a transit station are delivered to their doorsteps. On the return trip to the station, the vehicle boards new patrons in the manner of traditional, fixed-route, fixed-stop feeder-bus service. Continuum...

In this research, simulation is used to explore how a bus bypass segment, also called a queue jump, affects traffic on a signalized arterial. Residual queues form at the site’s critical bottleneck and expand to street links upstream. A bypass, when designed to serve a bus stop as per AASHTO guidelines, is shown to reduce bus delays—if the stop resides on a congested link that is upstream of the one feeding traffic to the critical bottleneck. In contrast, using a bypass for a bus stop located immediately upstream of the critical bottleneck starves that bottleneck of flow. The damage thus...